Perceptions and Realities: Investigating Barriers to Artificial Intelligence (AI) Integration Among Medical Imaging Professionals within the United States By Adam Fisher Codi Harbour Katlyn Kiger Michael Oveson Kimberly Poss Emilee Scott A thesis submitted to the School of Radiologic Sciences in collaboration with a research agenda team In partial fulfillment of the requirements for the degree of MASTER OF SCIENCE IN RADIOLOGIC SCIENCES (MSRS) WEBER STATE UNIVERSITY Ogden, Utah August 13, 2025 2 THE WEBER STATE UNIVERSITY GRADUATE SCHOOL SUPERVISORY COMMITTEE APPROVAL of a thesis submitted by Adam Fisher Codi Harbour Katlyn Kiger Michael Oveson Kimberly Poss Emilee Scott This thesis has been read by each member of the following supervisory committee and by majority vote found to be satisfactory. ______________________________ Dr. Robert Walker, PhD MSRS Program Director, School of Radiologic Sciences ______________________________ Dr. Tanya Nolan, EdD Chair, School of Radiologic Sciences 3 THE WEBER STATE UNIVERSITY GRADUATE SCHOOL RESEARCH AGENDA STUDENT APPROVAL of a thesis submitted by Adam Fisher Codi Harbour Katlyn Kiger Michael Oveson Kimberly Poss Emilee Scott This thesis has been read by each member of the student research agenda committee and by majority vote found to be satisfactory. Date August 13, 2025 ______________________ August 13, 2025 ______________________ August 13, 2025 ______________________ August 13, 2025 ______________________ August 13, 2025 ______________________ August 13, 2025 ______________________ ____________________________________ Adam Fisher ____________________________________ Codi Harbour ____________________________________ Katlyn Kiger ____________________________________ Michael Oveson ____________________________________ Kimberly Poss ____________________________________ Emilee Scott 4 Abstract Artificial intelligence (AI) is an ever-evolving factor in today’s world, especially within the radiology department of healthcare. While AI can be seen with positive results and adequate usage around the globe, there seems to be little known or recorded usage within the United States. To further investigate why this might be, quantitative plans were made to evaluate various radiology personnel and their perceptions about AI. A Likert-style survey was built and questioned participants about their perceptions regarding AI usage, potential barriers of AI implementation, associated AI risks and benefits, as well as various demographic information. The surveys were sent out through social media outlets and through emails through a convenient sampling fashion. There were 253 surveys collected back from various regions of the United States, and were filled out by technologists, managers, radiologists and radiologist extenders. The data from the surveys were then cleaned up and evaluated through one-way ANOVA and correlation statistics to check for significant findings between the research questions and the variables. There was a significant difference between facility size and the sum of AI usage (p=<.001). Significant findings were found in the overall sum of barriers based on age of participants (p=<.001). There was a significant difference between age and benefits (p=<.001). There was a significant difference between age and risks (p=<.001). The research showed that there was a limited use of AI across the United States, and that there were several barriers identified associated with factors, such as age, years worked, facility size and roles within radiology departments. Future research should focus on reaching larger sampling sizes, seek to avoid further convenience sampling, and focus on getting more radiology departments better educated on current AI systems and their associated benefits. 5 Acknowledgements We would like to express our most sincere gratitude to all of those who have supported, tolerated, and guided us throughout this research project. First and foremost, we would like to thank Dr. Tanya Nolan, whose experience and expertise were instrumental in the compilation of our paper. We would also like to recognize Dr. Taylor Ward and Dr. Laurie Coburn for their teachings, insightful direction, and feedback. We would lastly like to thank Cathy Wells for answering endless emails with patience and kindness and for guiding our educational journeys. We would like to thank the faculty and staff of the Dr. Ezekial R. Dumke Department of Radiological Sciences at Weber State University for aiding us in our goals and pushing us to achieve our educational dreams. With your help and through our achievements we know we can improve the future of radiological sciences. Finally, we extend our deepest appreciation to our family, friends, work, hospitals, and clinics. Whose understanding, patience, and encouragement made this journey possible. Adam-To acknowledge my fellow peers and friends, my dearest wife and family for continually supporting me through and through, as well as my work team for working with my schedule and making this experience as smooth and beneficial as possible. Codi- I would like to acknowledge my classmates, coworkers, family and friends for guiding and supporting me on this educational journey. Thank you for the kind words of encouragement along the way. A special thank you to my husband for the unwavering support and to my parents for instilling the importance of lifelong learning. Katlyn- To acknowledge my peers, coworkers, family, friends and fiancé for supporting me throughout this entire educational career and research project. Michael- To acknowledge my fellow classmates, colleagues, wife, family and friends. Thank you for your encouragement, love, kindness, and support. Kimberly- I would like to acknowledge all the researchers in this group for all their support along with my husband and kids who were patient when I needed them to be to achieve my educational goals. Thank you to Cathy Wells for all her guidance and patience throughout the entrance process into the master’s degree program. Emilee- I would like to acknowledge my classmates, professors, my wonderful husband, family, and friends. Thank you for all of your encouragement and support that has helped me throughout this experience. 6 Table of Contents Chapter 1: Introduction ....................................................................................................... 1 Statement of the Problem .................................................................................................... 2 Purpose of the Study ........................................................................................................... 2 Research Questions ............................................................................................................. 3 Nature of the Study ............................................................................................................. 4 Significance of the Study .................................................................................................... 5 Definition of Key Terms ..................................................................................................... 5 Summary ............................................................................................................................. 9 Chapter 2: Literature Review ............................................................................................ 10 The Use Of AI ................................................................................................................... 13 Barriers to AI Use ............................................................................................................. 15 Funding/Money ................................................................................................................. 15 Training ............................................................................................................................. 15 Resistance to Change ........................................................................................................ 16 Lack of Standardization .................................................................................................... 17 Protection of Patient Information...................................................................................... 18 Summary ........................................................................................................................... 19 Chapter 3: Research Method ............................................................................................. 20 Research Methods and Design(s)...................................................................................... 21 Population ......................................................................................................................... 22 Sample............................................................................................................................... 23 Materials/Instruments ....................................................................................................... 24 Operational Definition of Variables .................................................................................. 25 Survey Instrument ............................................................................................................. 28 Assumptions ...................................................................................................................... 28 Limitations ........................................................................................................................ 29 Delimitations ..................................................................................................................... 30 Ethical Assurances ............................................................................................................ 31 Summary ........................................................................................................................... 31 Chapter 4: Findings ........................................................................................................... 32 Usage................................................................................................................................. 33 Barriers .............................................................................................................................. 40 Benefits and Risks............................................................................................................. 42 Evaluation of Findings ...................................................................................................... 45 Summary ........................................................................................................................... 47 Chapter 5: Implications, Recommendations, and Conclusions ........................................ 48 Implications....................................................................................................................... 48 7 Recommendations ............................................................................................................. 49 Conclusions ....................................................................................................................... 50 References ......................................................................................................................... 51 Appendices ........................................................................................................................ 55 Appendix A: IRB Approval Letter ................................................................................... 56 Appendix B: Survey .......................................................................................................... 57 Appendix C: Examples of Participant Invitations ............................................................ 65 Appendix D: Tables .......................................................................................................... 68 Appendix E: Figures ......................................................................................................... 72 8 List of Tables Table 1. Descriptive Statistics Usage ............................................................................... 34 Table 2. Mean Usage by Facility Size .............................................................................. 36 Table 3. Mean Usage by Years Worked ............................................................................ 38 Table 4. Mean Usage by Age Group ................................................................................. 39 Table 5. Barriers Correlations ......................................................................................... 41 Table 6. Benefits Correlations .......................................................................................... 43 Table 7. Risks Correlations............................................................................................... 44 Table 8. Facility Size Descriptives.................................................................................... 67 Table 9. Current Professional Role Descriptives ............................................................. 68 Table 10. Years Worked Descriptives ............................................................................... 69 Table 11. Age Descriptives ............................................................................................... 70 9 List of Figures Figure 1. Current Professional Role Within the Radiologic Sciences .............................. 35 Figure 2. Facility Size ....................................................................................................... 37 Figure 3. Participants by State.......................................................................................... 38 Figure 4. Number of Years Worked................................................................................... 71 Figure 5. Professional Role............................................................................................... 71 Figure 6. Age of Participants ............................................................................................ 72 1 AI Use Within Radiology Chapter 1: Introduction Professional occupations of all kinds are experiencing technological advancements and changes that are enhancing efficiency, simplifying processes, and changing how problems are approached and dealt with. This is, perhaps, in many ways especially true of healthcare systems, and even more true within the field of radiology. Mohammad et al. (2018) found one of the biggest drivers of this evolution to be the everevolving status and use of Artificial Intelligence (AI). They defined AI as any program or software capable of learning, being taught, and capable of carrying out tasks under a set of given parameters, in ways that resemble human ways of accomplishing them. AI was conceptualized as early as the 1960s, but due to technology being locked behind the times no real progress was made for some decades. It wasn’t until the 1980s where significant advancements were made, largely in part to artificial neural networks and the learning systems they were capable of producing. One of the first AI products of this era was computer aided detection (CAD), software capable of being trained to find patterns in images that might otherwise be missed by human eyes (Mohammad et al, 2018). This CAD software, first adopted by the mammography department, is where the field of radiology really took its first big steps into the world of AI (Mohammad et al, 2018). Moving forward to within the last decade or so, more imaging modalities have adopted CAD systems to help survey and prioritize critical finds, such in the cases of musculoskeletal ultrasound scans and chest scans through standard x-ray and computed tomography (CT), that might otherwise be missed when being interpreted by a radiologist, for any number of reasons (Shin et al, 2021;Vimalesvaran et al, 2024). While AI systems, like CAD, have been helping with better outcomes for a number of different 2 scans, other AI applications are still being polished and refined, and showing a lot of potential for radiology to be able to further patient care. Statement of the Problem A problem with the introduction and utilization of AI in healthcare is a lack of understanding of which AI resources are available in the U.S. and the perceived beliefs around AI amongst technologists, radiologist extenders, radiologists and managers. There are many documented benefits associated with the use of AI, and there are likely several more advantages waiting to be discovered and studied. Outside the United States, there are a large number of countries utilizing AI in everyday radiology practices, and because of the frequency of use, these countries have adopted multiple laws and regulations that help guide the usage of AI. In the United States, there is still a lack of healthcare facilities utilizing AI in radiology and medical imaging (Ungureanu et al., 2025). Thus, healthcare departments are not reaping the benefits of AI including efficiency in reading exams, shorter turnaround times, better understanding of report interpretations in layman terms, quicker access to reports for patients and providers, and better workflows in departments to help technologists (Rudolph et al., 2024; Zech et al., 2024; Sieber et al., 2024). Purpose of the Study The purpose of this quantitative study is to research the perceptions and challenges related to an under-utilization of AI in healthcare systems. Because of misunderstanding, mistrust, and lack of resources, AI is an ineffective tool in assisting with workplace practices, productivity, and patient care. Research questions highlight the current usage of AI in the United States, the challenges and benefits of using AI, and the current perceptions of healthcare professionals in imaging. Variables include perceptions 3 and behaviors of technologists, managers, radiologist extenders and radiologists in relation to AI resources and usage. A literature review was conducted comparing grounded theories of past researchers via a PRISMA method as foundation to the study. Researchers have developed a survey with closed and open-ended questions wherein working technologists, managers, radiologist extenders and radiologists within the U.S. may reflect upon their own perceptions, availability, competency, and use of AI technology. Research Questions This study aims to investigate the current usage of AI in medical imaging, the barriers to utilizing AI, the perceptions among medical imaging personnel (managers, technologists, radiologist extenders and radiologists), and the benefits or risks of AI implementation in a medical imaging department. A quantitative survey was conducted with imaging professionals working within imaging centers and hospitals across the United States. The research questions include: Q1. What is your current usage of AI in radiology? H10: There is no use of AI within radiology departments located within the United States. H1a: There is limited usage of AI within radiology departments located within the United States. Q2. What are the barriers to utilizing AI in radiology? H10: There are no barriers for the utilization of AI in radiology. H1a: There are several barriers for utilization of AI in radiology departments including, but not limited to, attitudes and behaviors of 4 healthcare professionals, scarce resources, and limited professional development and training. Q3. What are the perceived benefits and risks among technologists, managers, radiologist extenders and radiologists in implementing and utilizing AI in radiology? H10: There are neither benefits nor risks in implementing and utilizing AI in radiology departments. H1a: There are both significant benefits and risks in implementing and utilizing AI in radiology departments. Nature of the Study This study uses a quantitative design to explore the current state of AI usage in medical imaging across the U.S. Specifically, this study investigates the extent of AI implementation, the barriers that impede its utilization, and the perceived benefits and risks associated with AI integration among medical imaging professionals. Data was collected through a structured survey that was distributed among employed imaging personnel in the U.S. Data was collected from items measured on a Likert-style scale distributed via social media outlets. The minimum goal of the researchers was to receive 50 completed surveys for statistical analysis. The survey was composed of questions that aligned with the research topic. Most previous research has been completed outside of the U.S. by researchers who utilized similar methods and survey design. A Qualtrics survey was created by the researchers to assess multiple data related to the overarching independent and dependent variables as means to support accurate measures. This 5 ensured the instrument’s quality, integrity, organization and ease of distribution. Data was then analyzed, rated and presented in the following chapters. Significance of the Study The significance of this research is to ascertain the current level of knowledge of AI use in the United States. This will allow us to compare the possibilities of AI practice in the future to enable organizations to overcome challenges for implementations. With this research, we will be able to raise awareness of tools to overcome the challenges faced by technologists, radiologist extenders, radiologists and radiology managers in regard to utilizing AI as part of their daily workflow. By encouraging a better understanding of AI use, we promote its effectiveness in assisting with workplace practices, productivity and patient care. Definition of Key Terms Accuracy. Key performance metric in AI models that measures overall correctness. This metric helps assess AI reliability in radiology. (Hong et al., 2023). AI-Augmented Radiology. The concept that AI serves as an enhancement to radiologists rather than a replacement. AI systems assist in image interpretation, workflow management, and decision support, allowing radiologists to focus on complex cases and improve efficiency. (Geis et al., 2019). AI Competency. The level of knowledge and skills required for radiology professionals to implement and use AI effectively. This includes understanding AI tools, interpreting AI-generated results, and integrating AI into clinical workflows. (Jassar et al., 2025). Artificial intelligence (AI). A branch of computer science that enables machines to simulate human intelligence, including learning, reasoning, and problem-solving. In 6 radiology, AI is used for image analysis, diagnosis, workflow optimization, and predictive modeling. It encompasses techniques like machine learning (ML), deep learning (DL), and computer vision. (Geis et al., 2019). Automation in Imaging. The use of AI to streamline radiology workflows by automating repetitive tasks such as image acquisition, protocol selection, and report generation. This helps improve efficiency, reduce errors, and allow radiologists to focus on complex cases. (Wenderott et al., 2024). Benefits of AI in Radiology. AI offers various advantages, including increased diagnostic accuracy, faster image interpretation, reduced workload, cost savings, and enhanced workflow efficiency. It can assist in early disease detection, leading to better patient outcomes. (Khalifa & Albadawy, 2024). Blockchain. Digital information is stored in a digital block. Once that information is in that block it cannot be changed. An additional block with changes can be added via a link that connects to the previous block. This creates a virtual chain. All participating computers will copy this blockchain as blocks are added. The blockchain secures information by leaving a print of all who access and change the information. (Zuo, 2025). Challenges of AI in Radiology. Despite its benefits, AI adoption faces obstacles such as algorithm bias, data privacy concerns, regulatory hurdles, ethical considerations, and the need for extensive training and validation. Integration into existing radiology workflows also presents challenges. (Brady et al., 2024). Computer-Aided Detection (CADe). AI-powered tools that assist radiologists by flagging potential abnormalities in medical images, such as tumors, fractures, or lung 7 nodules. CADe is designed to highlight areas of concern but does not provide a definitive diagnosis. (Masud et al., 2019). Computer-Aided Diagnosis (CADx). AI-driven systems that go beyond detection by providing diagnostic suggestions based on imaging features. CADx aids radiologists in differentiating between benign and malignant lesions or classifying disease severity. (Yeasmin et al., 2024). Current Usage of AI. The ways AI is currently applied in radiology, including image interpretation, workflow optimization, clinical decision support, automated reporting, and predictive analytics. AI is used in multiple imaging modalities, including X-ray, CT, MRI, and ultrasound. (Khalifa & Albadawy, 2024). Data Privacy and Security. Refers to the protection of patient data used to train and deploy AI models. Ensuring compliance with regulations like HIPAA (Health Insurance Portability and Accountability Act) is critical to safeguarding patient confidentiality and preventing unauthorized access. (Brady et al., 2024). Deep Learning (DL). A subset of machine learning that uses artificial neural networks with multiple layers to analyze complex patterns in medical images. DL is particularly useful for tasks such as tumor detection, image segmentation, and anomaly detection. (Geis et al., 2019). Global AI Usage. The differences in AI adoption and implementation across various countries. Factors such as healthcare policies, infrastructure, regulatory approvals, and access to data influence how AI is used in radiology worldwide. (Khalifa & Albadawy, 2024). 8 Human-AI Collaboration. The partnership between radiologists and AI tools, where AI enhances, rather than replaces, human expertise. Radiologists interpret AI-generated insights while applying their clinical judgment to ensure accurate diagnoses and patient care. (Geis et al., 2019). Image Segmentation. An AI-driven process that divides a medical image into meaningful regions, such as isolating a tumor or highlighting anatomical structures. This technique helps in precise diagnosis and treatment planning. (Nair et al., 2025). Implementation of AI. The process of integrating AI into radiology departments, including selecting AI tools, training staff, updating IT infrastructure, and ensuring regulatory compliance. Successful implementation depends on institutional support and workflow integration. (Brady et al., 2024). Machine Learning (ML). A subset of AI where algorithms learn from large datasets to identify patterns and make predictions. ML is widely used in radiology for image recognition, disease classification, and anomaly detection. (Najjar, 2023). Natural Language Processing (NLP) in Radiology. AI methods that analyze and interpret radiology reports, clinical notes, and medical literature. NLP helps extract key findings, standardize reporting, and improve communication between healthcare providers. (Casey et al., 2021). Neural Networks. Computational models inspired by the human brain, used in deep learning to process and analyze medical images. Neural networks improve pattern recognition and enhance AI-driven diagnostic capabilities in radiology. (Najjar et al., 2023). 9 Perception of AI. The attitudes and beliefs of radiology professionals, including technologists, managers, and radiologists—toward AI. This includes concerns about job displacement, trust in AI-generated results, and the willingness to adopt AI-driven tools. (Brady et al., 2024). Precision. Key performance metric in AI models that indicates the proportion of correctly identified positive cases. This metric helps assess AI reliability in radiology. (Kocak et al., 2025). Recall. Key performance metric in AI models that measures how well the model identifies actual positives (its sensitivity). This metric helps assess AI reliability in radiology. (Kocak et al., 2025). Regulatory and Ethical Considerations. The legal and ethical frameworks governing AI use in radiology. This includes FDA approvals, HIPAA compliance, data privacy laws, and ethical concerns regarding AI bias, transparency, and accountability. (Brady et al., 2024). Sensitivity. Measure of AI accuracy in detecting diseases that refers to the ability to correctly identify diseased cases (also known as true positive rate). High sensitivity is essential for reliable AI models in radiology. (Klontzas et al., 2025). Specificity. Measure of AI accuracy in detecting diseases that refers to the ability to correctly classify healthy cases (also known as true negative rate). High specificity is essential for reliable AI models in radiology. (Klontzas et al., 2025). Summary Over the past few decades, the rapid advancement of artificial intelligence (AI) in the U.S. has generated a sense of unease among the public, largely due to limited 10 understanding of how AI is developed and applied. The purpose of this study is to examine the current use of AI within the field of radiology and to identify the associated benefits, risks, and barriers. This research aims to underscore the importance of enhancing knowledge and awareness of AI to support its effective and responsible integration into radiological practice. Chapter 2: Literature Review Artificial Intelligence (AI) has become a driving factor for change in many different businesses and settings, offering easier and often better ways to solve problems, improving efficiency, and saving time and money. Healthcare, radiology in particular, is among the leading industries that have gained a lot from AI and continues to show a vast amount of potential as AI continues to advance (Mohammad et. al, 2018). AI has come a long way from its humble conception in the 1960s till now (Mohammad et. al, 2018). Mohammad et. al (2018) explains that what started as math equations and simple programming would eventually produce programs and systems capable of learning, predicting, and pattern recognition. They further explained that these leaps and bounds in AI production took time to achieve, largely due to early ideas being hindered by technologies of the time. It wasn’t until the 1980s that technology finally started advancing in significant ways which allowed for advanced learning systems to be produced by products like artificial neural networks (Mohammad et. al, 2018). These artificial neural networks, patterned after actual human neuron systems, work very similarly to a human brain, and would eventually give rise to teaching techniques, such as deep learning (DL), that would change and shape how AI could be utilized (Mohammad et.al, 2018). 11 Among the greatest AI advancements of the 1980s and 90s were CAD systems, which allowed for pattern recognition in images, including very subtle changes or patterns that might be missed by the human eye, and in some cases, detecting patterns unseeable to the human eye (Mohammad et.al, 2018). The first radiologic department to utilize this technology was the mammography department, and it was very quickly established as a critical tool in improving radiologists’ rates of cancer detection (Mohammad et. al, 2018). After years of successfully implementing CAD systems into the Mammography workflow, other imaging modalities started taking note and would start to make the jump into the world of AI integration. Mohammad et. al (2018) explain that standard x-ray and CT have since integrated CAD software into practice and have been perfecting its use over the years to help find tricky lung nodules and lesions, including smaller earlier-stage malignancies that would typically be missed by even experienced radiologists. While CAD systems do differ from modality to modality in results and how well they perform, Mohammad et. al was quick to point out that some CT CAD systems have been able to detect as much as 70% of originally missed lung lesions by radiologists, solidifying CAD software’s importance as a radiologist’s diagnosing tool. Improving imaging equipment and software are not the only ways AI has been changing and benefiting the world of radiology. Language models, such as Chat GPT, have only gotten better over the recent years and continue to aid in diverse human-like communication and text, which has been a real aid for both radiologists and patients alike (Najjar R., 2023). Park et. al (2024) explains that some AI programs are utilizing Chat GPT in the reporting process by helping patients receive imaging results in a more timely fashion and even making them easier to understand, alleviating stress and creating a more 12 positive experience. Other programs are acting as screenings to help prioritize more important or severe cases for those in the emergency room, helping radiologists prioritize and report on image series with more at stake, such as in the cases of head and neck traumas discovered by CT scans (Vimalesvaran, K., 2024). While many of these various programs and functions are still being tested, and certainly not without flaws, it is clear that AI has created a foundation of high expectations and hope for enhanced radiographic performance. With all these advancements, one might ask why professionals and the public do not hear or see more about the use of AI here in the United States? While searching the literature, there are many sources of AI and technical advancements currently used in Europe and China, but data found for the U.S.is far more scarce. There are several factors that may be contributing to this. These include, but are not limited to, cost, efficacy of AI, and the perceptions of radiologists, imaging technologists, managers, and the general public (Strohm, Lea et. al, 2020). To investigate this further, a literature review matrix was created and consisted of peer-reviewed articles found through the Stewart Library search engine of Weber State University. Due to the scarcity of data found within the United States, sources from other countries were included in this process. Topics for the literature search included AI perception from radiologists and the public, current AI trends and technology, and current limitations and strategies. Searches were, however, limited to key words such as radiologists, technologists, AI, influence, perception, and patient care, and only included peer-reviewed articles. All articles used were limited to those published within the last 5-6 years, except in cases of older articles containing 13 historical significance and understanding to better understand and appreciate the advancements of AI. The Use of AI As of March 2025, there are more than 700 AI and ML FDA-approved medical devices for use in radiology in the U.S. (U.S. Food & Drug Administration, 2025). This number comes from a list on the US FDA website that monitors the active listings of AI & ML technologies currently being utilized. To date, only a few of the technologies are being reimbursed by insurance companies, but all these technologies are being used, both in the research setting and in clinical practice. Since 2019, the FDA has been working on a proposed regulatory framework for modifications to AI/ML-based software (Wichmann et al., 2020). It is estimated that the overall global market for AI in medical imaging will rise from $21.5 billion to $264.9 billion from 2018 to 2026 (Wichmann et al., 2020). In Europe, AI is most relevant in CT and MRI. According to a study conducted by the European Society of Radiology, AI tools help with clinical efficiency but still require radiologist oversight (Moreno Zanardo et al., 2024). Both inside and outside the U.S., many of the most successful implementations of AI have been completed at large academic institutions with large funding resources (Wichmann et al., 2020). Additionally, many of the AI/ML applications are being trained, validated, and tested in research settings before they are utilized in everyday clinical practice within imaging departments (Wichmann et al., 2020). AI is most commonly being used in medical imaging with brain, thoracic, abdominal, and pelvic imaging, colonoscopies, mammography, and radiation oncology 14 (Hosny et al., 2018). AI in brain imaging helps radiologists characterize abnormal and incidental findings, and it is often used to assess the stability of brain lesions on serial scans (Hosny et al., 2018). In high-risk patients, low-dose lung screens are helping radiologists to find lung nodules and assisting in determining whether they are benign or malignant (Hosny et al., 2018). One of the biggest impacts on medical imaging is the AI assistance of oncological staging and monitoring in patients with cancer. AI is being used to plan treatment for patients and help determine prognosis and success of treatments when comparing subsequent scans (Hosny et al., 2018). As it appears, there are several main areas in medical imaging that AI is regularly being used. Some of these areas include information management, quality assurance, diagnoses, and image post-processing. AI assistance in information management is being implemented to improve processes such as patient scheduling, technologist workflow automation, and report generation for ordering providers and patients. For quality assurance, AI helps technologists follow regulatory and imaging protocols, quickly and efficiently identify errors in images and on equipment and effectively receive peer reviewed information and data such as equipment manuals and troubleshooting guides. Assisting in diagnosing patients and the abilities to initiate image post-processing is where AI is shining the brightest in medical imaging. When diagnosing patients, radiologists and ordering providers are utilizing AI tools to help with early and small lesion detection, comparison of lesions through serial scanning, correlations with known pathologies and support in making clinical decisions. Post-processing may also include tools such as 3D reconstruction and anatomy subtraction on cross-sectional images, image enhancement, surgical pre-planning and segmentation. 15 Barriers to AI Use The barriers to utilizing AI in the U.S. include, but are not limited to, funding, training, and a resistance to technological changes with a lack of standardization and concerns for patient confidentiality and protection. Each challenge will be detailed within the next few sections. Funding/Money Current research has shown that AI will transform the world of medical imaging and diagnostics, but at a steep price. Adoption of AI applications in radiology is deemed quite difficult because mature healthcare systems use a prospective payment system for paying healthcare providers. Payments made under a prospective payment system for diagnostic radiology services are meant to cover the cost of conducting the imaging study and the associated reporting. Prospective payment systems are generally slow to account for increases in costs that are due to the adoption of innovations, which may not bear well for the adoption of AI in radiology (Lobig et al., 2023). The reimbursement system of healthcare has a duty to protect patients by ensuring that all physicians and treating practitioners have access to tools that can have potentially revolutionary benefits on patient care. With that principal focus, market leaders will likely lead the way in promoting and practicing medicine with the newest AI innovations, but global healthcare systems should be careful not to dive into a two-tiered system because while some of the healthcare systems can afford AI, others cannot (Ranschaert et al., 2019). Training The education and training of the healthcare workforce is a significant barrier to the adoption of AI. A lack of strong training plan is also a barrier to the integration and 16 utilization of AI in radiologic sciences. (Strohm et al. 2020) stated that radiographers believed that in order for AI to be successfully utilized, it must have a smooth transition to workflow practices. Such things that would be needed to appear seamless in transition are the flow into the ITS systems, such as the picture archiving and communication systems (PACSs). According to radiographers, the perceived transition of AI into the PACS system would be that the output of the AI application should be displayed with the least possible clicks. AI applications should be implemented without large changes to routines and workflow practices, such as avoiding additional steps for reporting the result of the AI application. An example is that the interviewed users experienced the integration of BoneXpert into the PACS as very smooth, being the main reason for its perceived user-friendliness. However, concerns remain about the integration of other AI applications into the PACSs (Strohm et al, 2020). Resistance to Change Physician ‘buy-in’ and engagement is a known barrier to the utilization of AI. Many physicians do not engage fully with a new AI process, and this lack of engagement may produce clinical problems. Additionally, ignorance or reluctance to AI could lead to poor patient outcomes. Without the proper buy-in from users and adequate engagement of clinical staff and healthcare leaders, even the best AI tools are unlikely to be accepted and integrated and will be unable to improve clinical outcomes. When a system is perceived by clinicians as threatening and interfering with autonomy and critical thinking, usage will be compromised. Consistent and determined effort will be essential to nurture clinicians’ engagement with the usage of AI in radiography. 17 Clinical staff may also demonstrate skepticism regarding the current capabilities for AI in particularly complex patient cases or in regard to conditions wherein management is known to be difficult and complex. Resistance to the use of AI in imaging modalities can often be correlated to workers being worried about job displacement. The fear of job displacement is consistently a theme for resistance to change and the utilization of new technology. Clinicians are not immune to the fear of unemployment or radical job changes arising from increasingly complex and competent AI systems (Nur Ahmed et al., 2023). This ideology leads to behaviors and practices designed to ensure consistent relevance for the roles of human practitioners in healthcare. Lack of Standardization Throughout different sites, multiple equipment vendors, and a variety of modalities cause barriers to utilizing AI in radiology. Standardization is a common regulating practice to govern technology and implementation harmonization (Jobin et al, 2019: 396) and is argued to be an important piece of the AI policy ecosystem (e.g., Shank, 2021). It becomes important to label studies commonly throughout the U.S. This helps make deep learning more comparable. For something to be measured against the same criteria, it does, however, require likeness amongst the entities evaluated (Busch, 2011, Bowker and Star, 1996). When terminology or labeling is inconsistent, communication between systems can be lacking. Standardization allows data to be labeled the same, so it is easy to locate, can be shared, and is easy to apply AI algorithms. An example in radiology is CT of the abdomen and pelvis, where it is commonly called in some sites CT body. With the different labels, it will cause inconsistencies when pulling data to train the AI. When striving for effective implementation of AI in 18 radiology, standardization is key for reliability, which in turn will benefit the radiologists and patients. Protection of Patient Information When developers of AI have access to large amounts of data, this can create opportunities for huge leaks of patient information (Ranschaert et al., 2019). Some questions may come up when protecting one's personal data such as, “If a patient’s data is used to develop applications sold for profit, are patients entitled to compensation?” and “Will informed consent be required only for patient data in the development of deeply annotated AI datasets?” (Ranschaert et al., 2019). One way to manage patient data is through transparency. Patients want to make sure their information will not be shared in ways they don’t understand (Ranschaert et al., 2019). So, as long as the patient understands, the more willing they are to opt in to sharing their personal data. The logic behind transparency, as a concept and metaphor, is that what can be seen can be known, and that we, by observation, are able to reach new insights (Ananny and Crawford, 2018). There are high expectations of what can be accomplished by AI transparency, but it is difficult to distinguish by which actions and practices the ideal is to be realized (Hogberg, 2024). New cryptography techniques can be used to keep patients' privacy. Blockchain methods use a database that continually updates blocks that contain a linked list of all previous transactions. The initial block is called the genesis block, which starts the blockchain. Each block following consists of a header and a block body. The blocks link together to form a chain in chronological sequence (Zuo, 2025). Any time anyone enters these blocks, a fingerprint is left to know who was using access to the block information. AI information can get corrupted if fed bad data, but with blockchain technology, the data 19 is tamper-proof, which in turn boosts trust in AI outcomes. This is probably the most sophisticated tool to protect patient information that we have at present. As noted, many barriers exist that affect the use of AI within radiology. Funding, training, staff resistance to change, lack of standardization, and the protection of patient information will hinder the growth of AI in radiology departments. The misconception that AI will replace the radiologist also keeps professionals from being “all in” to participate in growing AI and taking it to the next level. Summary As of today, the FDA has approved over 700 AI and ML medical devices for use in radiology in the U.S, with only a few being reimbursed by insurance. Despite the limited reimbursements, AI and ML technologies are used in clinical and research settings. Globally, the market for AI and ML in medical imaging is projected to grow significantly, especially in Europe, where these technologies are primarily used in CT and MRI. AI tools improve clinical efficiency but still require a technologist and a radiologist's oversight. AI and ML are most commonly applied to streamline workflows, ensure compliance with imaging protocols, and reduce errors while improving patient scheduling and generating reports more efficiently. Several barriers hinder AI and ML usage in radiology. Funding is a significant concern, as existing payment systems struggle to accommodate the cost of implementation. Training healthcare professionals and ensuring integration into workflows, like PACS, remains challenging. Resistance from clinicians, driven by fears of job displacement and disruption of autonomy, is common in many radiology departments. Finally, data privacy, standardization issues in study labeling, and patient 20 consent are critical concerns. However, despite these challenges, thoughtful implementation could lead to a more robust and equitable use of AI and ML in radiology. Chapter 3: Research Method AI is beginning to change how healthcare is delivered, especially in the radiology department. AI carries with it the inherent benefits that affect all staff, from managers down to the imaging technologists, and include benefits such as patient management, increased patient care and satisfaction, and increased department efficiency (Park et. al, 2024). In countries like China and throughout Europe, AI is being implemented and thriving, showcasing many benefits, such as easier to understand reports and quicker access to them for the patients; These benefits extend to both patients and radiologic staff, but they currently appear to be underwhelmingly utilized within the U.S. (Ungureanu et al., 2025). A problem with the introduction and utilization of AI in healthcare is a lack of understanding of which AI resources are available in the U.S. and the perceived beliefs around AI amongst technologists, radiologists, radiologist extenders and managers. The purpose of this quantitative study is to research the perceptions and challenges related to an under-utilization of AI in healthcare systems. Our research questions are as follows: (1) What is your current usage of AI in radiology?, (2) What are the barriers to utilizing AI in radiology, and (3) What are the perceived benefits and risks among technologists, managers, radiologist extenders and radiologists in implementing and utilizing AI in radiology? In order to explore these questions and better understand the perceptions of AI, a survey was constructed using a Likert scale design approved through an IRB board. The survey was built and powered by Qualtrics and was available for participants to take 21 under two major criteria: 1) a participant must be currently working within a radiology department, and 2) participants must currently be employed within the United States. The content of the survey extrapolates the research questions by exploring concepts that include participants’ current usage of AI within their departments and perceptions, barriers to implementation and benefits identification, perceived risks of AI implementation; data from the survey also included collecting demographic information from categories including age, region (state) of employment, and primary responsibly within the department (manager, radiologist, radiologist extender, technologist (which modalities). The Survey was distributed to a convenience sample and made available to medical imaging professionals through social media outlets, email invites, text, and other easily communicable electronic means. Data of a minimum of 50 participants was collected and analyzed. Research Methods and Design(s) The primary focus of the study is to explore the current usage of AI in medical imaging across the U.S. The method and design were selected for their alignment with the objectives to measure, quantify, and analyze perceptions and experiences across a broad participant base in a structured, statistically interpretable manner. For this reason, a quantitative research method is used with a descriptive and correlational design, complemented by deductive qualitative analysis. The quantitative research method was chosen given the study’s purpose: to measure variables such as the extent of AI implementation, perceived benefits and risks, and barriers to using AI in medical imaging. This method is ideal for identifying trends, making generalizations based on numerical data, and allows for objective reproducibility, facilitating future research on 22 the topic of AI in medical imaging. The descriptive design aims to help the study profile AI usage within medical imaging inside the U.S. It assists in identifying concerning aspects: adoption levels, barriers, benefits, and risks associated with AI. Correlation is used to examine the relationship between dependent variables, (usage of AI in medical imaging) and independent variables (measured barriers, demographic features, perception of risk, and perceptions of benefits). This allows for the identification of patterns and associations, supporting an understanding of how AI is perceived and used in medical imaging. One example of the dependent-independent variable association in this study is the relationship between AI usage in medical imaging (dependent variable) and concerns about AI accuracy and reliability (independent variable). This relationship highlights how perceptions of technological trustworthiness may influence AI implementation decisions. Additional correlations and comparative analyses are presented in appendix A, which details associations between the dependent variable, AI usage in medical imaging, and a range of independent variables, including measured barriers, demographic characteristics, and perceived risks and benefits. This study seeks to evaluate the current use of AI in medical imaging. Based on the correlations and comparisons noted in Appendix A, AI is not being employed in medical imaging departments across the U.S. as much as it could be. However, additional correlations can be drawn, and further research is necessary to compare AI utilization in medical imaging. Population The population for this study consisted of approximately 253 technologists, imaging managers, radiologist extenders (RA, RPA, PA) and/or radiologists/physicians currently employed within the field of radiologic sciences within the U.S. This 23 population is integral to the study, as the most relevant and insightful data will be obtained from professionals currently active in the field. This study seeks to explore imaging technologists' perceptions of how AI is influencing their professional roles, including whether it is contributing to increased stress due to additional training requirements and the growing reliance on computerized equipment. The data will also help explore whether the cost and maintenance required to use AI is perceived to be cost effective and resource heavy. As a result, the outcomes of this study may assist imaging managers and technologists to make choices relevant to AI and productivity when faced with staffing shortages. For radiologist extenders and radiologists, this study may assist with workflow and diagnostic options. As a result, this population is key to the study because the role of each professional will be changed, influenced, and evolved with the introduction of AI. Sample The researchers utilized a random convenience sample of imaging professionals located within regions of the U.S. close in proximity to the research team, including Montana, New Jersey, North Carolina, South Carolina and Utah. To reach the statistical G*Power required for this quantitative study, a minimum of 50 participants were desired to calculate correlations and comparisons. In reaching a yield of 253 responses, statistical analysis may be more meaningful and generalizable in understanding the current use of AI within the U.S. A higher statistical power helps distinguish a real or true effect of AI use in the U.S. that is not simply based on chance. Eligible participants were invited to the survey via social media, email lists within organizations, medical imaging colleagues and electronic communications, such as text 24 messages. Participants voluntarily completed the survey via the link shared by the group conducting the survey. Please see Appendix C for a list of examples of emails, social media blasts and a copy of the QR code used inviting participants to complete the survey. In order to be included in the study, participants must be currently employed in radiologic sciences in the U.S. Materials/Instruments The problem identified with the introduction and utilization of AI in healthcare is a lack of understanding of which AI resources are available in the U.S. and the perceived beliefs around AI amongst technologists, radiologist extenders, radiologists and managers. The quantitative survey purpose was to research the perceptions and challenges related to an under-utilization of AI in healthcare systems. The reasons being because of misunderstanding, mistrust, and lack of resources, AI is an ineffective tool in assisting with workplace practices, productivity, and patient care. The barriers that were identified are funding, training, resistance to change, lack of standardization, and protection of patient information. The survey was created to reach the population of technologists, imaging managers, radiologists, and radiologist extenders asking three questions. First, to what extent do imaging professionals currently use AI tools in information management, diagnoses, post-image processing, and quality assurance? Frequency of use was measured on a Likert scale from 0 never to 10 always. Second, to what extent did imaging professionals agree or disagree with statements referring to barriers related to adopting AI in a radiology practice. Barriers included funding, training, resistance to change, lack of standardization, and protection of patient information. Finally, the survey asked each 25 person to indicate their level of agreement with statements highlighting either the benefits or risks of AI use in radiology. Each question of the survey was composed of multiple positive or negative statements, to reduce bias or survey fatigue, that could be ranked on their level of agreement from 0 to 5. The data collected from this survey will be used to support or reject the researchers’ hypothesis regarding usage of AI in healthcare as correlated with is a lack of understanding of which AI resources are available in the U.S. and the perceived beliefs around AI amongst technologists, radiologists, radiologist extenders and managers perhaps related to barriers of funding, training, resistance to change, lack of standardization, and protection of patient information. Alternately, there will be no relationship or recognition of clear barriers that hinder AI usage. Operational Definition of Variables AI usage is the dependent variable in this quantitative study, and independent variables include perceived barriers, benefits, and risks. Each variable has been measured on a Likert scale utilizing several statements representing values associated with each variable. Detailed descriptions of each variable are listed below. AI Usage The frequency of AI use among imaging professionals during day-to-day tasks was measured using a scale from 0-10. Levels of use were grouped as: Never used (0-1), rarely used (2-3), sometimes used (4-6), often used (7-8), and always used (9-10). Areas of use were categorized as: ● Information management including workflow automation, patient scheduling and report generating such as MyChart. 26 ● Diagnosis such as lesion detection, image interpretation, and/or clinical decision support. ● Post-Image Processing including segmentation, 3D reconstruction, image enhancement. ● Quality Assurance such as error detection, protocol compliance and/or peer review support. Barriers to Using AI It was hypothesized that AI usage would be affected by perceived barriers in incorporating AI in radiology practices. A scale was developed from 1 (strongly disagree) to 5 (strongly agree) with 3 being (neither agree nor disagree). Statements incorporating common themes supported by literature perceived as barriers included: questions about accuracy and reliability, funding, protection of patient information, professional training, and a resistance to change. An aggregate score representing perceived challenges also provides insight into the samples’ understanding of AI and the feasibility of its use in practice. Benefits and Risks of AI Questions were asked to assess if there are benefits of using AI. Questions included: Do radiologic professionals believe there is improved diagnostic accuracy with AI use? Does it speed up interpretation of images and if it assists in quality control along with protocol adherence? Would AI help less experienced staff, enhance workflow efficiency across departments, while improving patient care and communication? Regarding the risks, radiologic professionals were asked whether they felt their skills would degrade with long term use. Would AI produce errors that are difficult to detect? 27 Did they feel AI would reduce their job opportunities, and does utilizing AI lack accountability with ethical and legal concerns? The research team incorporated the same scale used to measure barriers of using AI, 1 (strongly disagree) - 5 (strongly agree) with 3 being neither disagree or agree. With these measurements, the current state of what radiologic professionals perceive, whether positively or negatively, may correlate with usage and affect the dependent variable. Overall, the results obtained from the survey will indicate the amount of AI usage in the radiology field among a sample of professionals within the United States. AI usage is hypothesized to be influenced by independent variables measured on a Likert scale. The research team hypothesizes that when values for barriers are measured as a high value, AI usage will be low value. Conversely, if the independent variable of benefits has a high value, it is hypothesized that AI usage will also have a high value. Risks, like challenges, will have an indirect effect on AI usage. When risks are a high value, AI will not be utilized at high frequency. Knowledge was gained through demographic questions completed by volunteers that may give insight on professionals more or less likely to incorporate AI into their practice. Researchers gathered data on age, years in the radiology field, radiology profession, radiology specialties, size of location respondent work at, and what state were taken into consideration. With these measurements, the researchers will test their hypothesis as a means of characterizing if AI use in healthcare is related to a lack of understanding and faces unique barriers including funding, training, resistance to change, lack of standardization and protection of patient information. Bringing awareness to technologists, imaging 28 managers, physician extenders (RA, RPA, PA), radiologists/physicians may support continued growth in the profession’s knowledge base when it comes to AI use in the many facets that do and will affect the radiology and medical imaging departments. Survey Instrument English-only surveys were distributed across the U.S. via social media over a twoweek period. Over 260 responses were analyzed, with surveys from participants who did not consent to participate being removed. The 253 surveys were reviewed for missing information. This means that if the survey was incomplete, had partial information, or the participant did not respond, a zero score was used for analysis. Survey questions were then grouped based on the usage of AI in imaging departments, perceived barriers, benefits, and risks of using AI. One-way ANOVA analysis tests with correlation and descriptive analysis were then run using SPSS and compared to independent variables such as age, how many years the individual has worked in radiological sciences, and what current role they hold within radiological sciences. Researchers observed a significant presence of technologists vs non-technologists including managers, radiologist extenders, and radiologists. Questions were asked by researchers inquiring about the current usage of AI in medical imaging by these two groups. Assumptions In research, assumptions are beliefs or conditions accepted as true without direct evidence, serving as foundational elements that guide the study’s design, implementation, and interpretation of results. They are important because they clarify the scope of the research, set expectations for data collection and analysis, and help readers understand 29 the context in which findings are presented. In this study, several assumptions were made to support the validity and relevance of the results. ● Participant Honesty: It is assumed that participants provided truthful and thoughtful answers based on their professional experiences and perceptions of AI in radiology. ● Comprehension of Survey Items: It is assumed that participants fully understood the survey questions and Likert-scale items related to AI usage, benefits, risks, and barriers. ● Relevance of Perception: It is assumed that professional perception influences AI integration in radiologic practices, and that measuring perception provides meaningful insight into broader systemic issues. ● Technological Accessibility: It is assumed that participants had sufficient access to technology (internet, digital devices etc.) to participate in the survey. ● Sample Representativeness: While convenience sampling was used, it is assumed the sample provides a reasonable cross-section of medical imaging professionals in the U.S. for the purposes of exploring trends. Limitations This study has several limitations that should be considered when interpreting the results. These include convenience sampling, self-reporting bias, a lack of longitudinal data, and geographic imbalance. The use of a non-random, convenience sample may limit the generalizability of the findings to all radiologic professionals in the United States. Additionally, because the data were self-reported, there is potential for bias if respondents overstated or understated their perceptions or usage of artificial intelligence. The cross- 30 sectional nature of the study also poses a limitation, as it captures perceptions at only one point in time and does not account for evolving views or trends. Furthermore, despite efforts to ensure geographic diversity, responses may still be concentrated in specific regions, which could affect the representativeness of the data. To help mitigate these issues, a larger-than-required sample size (n = 253) was collected to enhance statistical power and provide a more robust analysis. Delimitations This study included several key delimitations to maintain focus and alignment with its stated purpose and available resources. First, the research was limited to radiologic professionals currently employed in the United States, intentionally excluding international perspectives. The study also utilized a strictly quantitative descriptive and correlational design, omitting qualitative methods such as interviews or focus groups that might have provided greater depth. Participation was limited to specific professional roles - technologists, radiologists, radiologist extenders, and imaging managers - excluding other relevant stakeholders like IT staff, administrators, or AI vendors. Additionally, the survey focused only on four domains of AI use: information management, diagnosis, post-processing, and quality assurance, rather than exploring the full range of AI applications in healthcare. Finally, the study assessed participants’ perceptions of AI usage rather than actual usage data or performance outcomes. These choices helped ensure a manageable scope and coherent direction for the research. Ethical Assurances Prior to data collection, ethical approval was obtained from the Weber State University Institutional Review Board (IRB), and the study was conducted in accordance 31 with all relevant ethical standards concerning participant privacy, informed consent, and data security. All participants were provided with a digital consent form that clearly explained the study’s purpose, the voluntary nature of participation, the anonymity of responses, and the right to withdraw at any time. To ensure confidentiality, no identifying personal or professional information was collected, and all data was securely stored in password-protected files accessible only to the approved research team. Survey data was used solely for the purposes of this research and will not be shared beyond the academic context without explicit permission. All data collection and handling procedures were fully compliant with IRB guidelines and HIPAA standards, where applicable. Summary AI is transforming healthcare delivery, particularly within radiology. While countries like China and those in Europe have seen success in AI integration, its utilization in the U.S. remains limited. This study utilized a quantitative, descriptive, and correlational research design to investigate the current usage, perceived benefits and risks, and barriers to integrating AI in radiology departments across the United States. A structured Likert-scale survey was designed and distributed via Qualtrics to a convenience sample of imaging professionals - including technologists, imaging managers, radiologist extenders, and radiologists - actively working within the U.S. The survey explored AI usage (the dependent variable) in four key domains: information management, diagnosis, post-image processing, and quality assurance. It also examined three major categories of independent variables: barriers (e.g., funding, training, standardization, resistance to change, and data protection), benefits (e.g., 32 improved workflow efficiency, accuracy, and patient communication), and risks (e.g., skill degradation, job displacement, and ethical/legal concerns). A total of 253 participants responded, exceeding the minimum required sample size for statistical power. Data was collected anonymously and analyzed using descriptive statistics and correlational methods (e.g., ANOVA) to determine associations between perception variables and AI adoption rates. The data analysis will help determine whether factors such as mistrust, misunderstanding, or lack of institutional support are inhibiting the adoption of AI, despite its availability and potential benefits. The population sample was intentionally limited to professionals in radiologic sciences within the U.S. to better understand the national landscape and inform future educational, operational, and technological strategies in domestic healthcare settings. Ethical protocols including informed consent, confidentiality, and IRB approval were followed throughout the study. Chapter 4: Findings The purpose of the research presented is to show if AI is being used in radiology departments across the U.S. This chapter delves into four areas of research results of AI; usage, barriers, benefits and risks. The research documents which age groups, years participants worked in the field and facility size are using AI the most and least. One-way ANOVA and descriptive statistics were computed to see the significance of use between these groups. While looking at the barriers, benefits and risks for AI, One-way ANOVA was configured to compare the age of participants, years worked and professional roles. Results follow along with visual figures and tables within this chapter below. 33 Usage The first research question stated, “What is your current usage of AI in radiology?” with null and alternative hypotheses being either there is no use of AI within radiology departments located within the U.S. or there is limited usage of AI within radiology departments located within the U.S. Overall, every one way ANOVA statistics ran all supported that the p-value of the usage questions compared to age, years worked, and facility size was significant therefore rejecting our null hypothesis emphasizing that there is limited usage of AI within the U.S. The current knowledge of AI in the U.S. among radiology professionals was split into two groups due to the volume of some subgroups. The non-technologist group incorporated radiologists, radiologist extenders and managers. The technologist group included all other responses. Using a scale from 010, the average knowledge of use of AI in radiology for the non-technologist group was 4.03 and for the technologist group was 3.2. This average was an aggregated score from four categories to define the areas in which non-technologists and technologists in radiology utilize present day AI. The four categories are information management, diagnosis of images, post-processing of images and quality assurance. Of these four groups of AI usage, based on the non-technologist and the technologist combined, the dependent variable displayed positively skewed and kurtosis results in most categories. Results for each show information management positively skewed (skew = 1.913, SEskew = 0.153), positive kurtosis (kurtosis = 2.724, SEkurtosis = 0.305), diagnosis of images positively skewed (skew = 1.912, SEskew = 0.305), positive kurtosis (kurtosis = 2.675, SEkurtosis = 0.305), post-processing of images positively skewed (skew = 1.108, SEskew = 0.153), negative kurtosis (kurtosis = -0.211, SEkurtosis = 0.305) and 34 quality assurance positively skewed (skew = 2.091, SEskew = 0.153), positive kurtosis (kurtosis = 3.495, SEkurtosis = .305) see Table 1. Table 1 Descriptive Statistics Usage N M SD skewness kurtosis Usage - Information Management 253 1.36 2.51 1.913 2.724 Usage - Diagnoses 253 1.36 2.52 1.912 2.675 Usage - Post Image Processing 253 2.32 3.31 1.11 -0.211 Usage - Quality Assurance 253 1.27 1.27 2.09 3.495 In each group a bar graph was also created to show the comparison of technologists versus non technologists who participated in the survey and data shows that more technologists answered by nearly 100 participants see Figure 1. 35 Figure 1 Current Professional Role Within the Radiologic Sciences A one-way ANOVA statistical analysis showed that there was a significant difference in usage based on the size of the facility that participants currently work at. This supports a rejection of our null hypothesis that there is no usage in radiology departments located within the U.S. and that size of facility does make a difference in the amount of AI usage in those related departments. There was a significant difference between facility size and the sum of usage (F34,218 = 19.179, p = < .001) and about 31% of the variance in usage was explained by the size of the facility (ω2 = 0.313). In the descriptive statistics run on the usage questions in comparison to facility size question one with usage related to information management the mean of 2.7 in the “other” option areas of radiography uses information management the most. For AI usage in diagnosing larger hospitals use this the most at a mean of 2.6. Post processing people who worked in larger hospitals also said they used this most often with a mean of 3.8 and lastly AI usage in quality assurance large hospitals and physician's office or clinic both 36 had the same mean of 2.25 reflecting that they both use quality assurance the most in comparison to the other facility sizes see Table 2. Table 2 Mean Usage by Facility Size Information Diagnoses Management Post Image Quality Processing Assurance N M M M M Missing Information (0) 91 0.29 0.30 0.65 0.43 Outpatient Facility or Imaging 17 1.06 1.24 1.59 1.29 Rural Hospital 1-25 Beds 9 2.67 1.56 2.67 1.00 Small Hospital 25-100 Beds 7 2.29 2.00 2.14 0.14 Medium Hospital 100-499 Beds 78 1.86 1.82 3.49 1.81 Large Hospital >500 Beds 40 2.25 2.63 3.85 2.25 Physician’s Office or Clinic 4 1.75 1.75 3.5 2.25 Other 7 2.71 2.00 3.29 1.43 Total 253 1.36 1.36 2.32 1.27 Center 37 From the one-way ANOVA statistics, an interesting visualization for the majority of the participants showed that most worked in medium sized hospitals that averaged in 100-499 beds (N = 78) see Figure 2. Figure 2 Facility Size Another interesting visualization based on the participants in the survey shows the states that each person listed if and when they completed the survey in its entirety. There were 33 different states that contributed, many with multiple participants. The majority of participants were from Utah (N = 64), second New Jersey (N = 23) and third North Carolina (N = 9) being the highest participating states, see Figure 3. 38 Figure 3 Participants by State Descriptive analysis was run for usage and years worked in the radiology field. Results showed personnel with 25 plus years of experience used AI the most with a mean of 12.037 and the least being 11-15 years in the radiology field with a mean of 6.227 see Table 3. Table 3 Mean Usage by Years Worked N M 0 91 1.66 5 Years and Under 51 7.61 6-10 Years 32 8.81 39 11-15 Years 22 6.23 16-20 Years 18 9.94 21-25 Years 12 11.33 25+ Years 27 12.04 Total 253 6.32 Descriptive values were obtained for age group and their use of AI. Age groups of 48-57 are using AI the most with a mean of 10.764 and the least being 28-37 age group who completed the survey with a mean of 7.849, see Table 4. Table 4 Mean Usage by Age Group N M 0 92 1.64 18-27 34 7.91 28-37 46 7.85 38-47 36 9.44 40 48-57 34 10.76 58+ 9 10.56 Prefer not to answer 2 8.00 Total 253 6.32 Barriers Research question two stated, “What are the barriers to utilizing AI in radiology?”. This question looked to test whether or not there are barriers to AI use in radiology departments in the US. Some of these barriers are attitudes and behaviors of healthcare professionals, scarce resources and limited professional development and training. The barrier questions were answered using a scale of 1-5 with 1 being “Strongly Disagree”, 2 being “Disagree”, 3 being “Neither Agree nor Disagree”, 4 being “Agree” and 5 being “Strongly Agree”. A one-way ANOVA statistical analysis showed that there are significant findings in the overall sum of barriers based on age of participants, years worked, and current professional role. This finding correlates with the rejection of the null hypothesis that there are no barriers for the utilization of AI in radiology. There was a significant difference between age and barriers (F17,235 = 12.063, p = < .001) and about 45% of the variance in barriers was explained by the age of participants (ω2 = 0.451). There was a significant difference between years worked and barriers (F17,235 = 8.920, p = < .001) and about 38% of the variance in barriers was explained by the number of years worked (ω2 = 41 0.376). There was a significant difference between current role and barriers (F17,235 = 9.842, p = < .001) and about 41% of the variance in barriers was explained by current professional role (ω2 = 0.407). The correlation between barriers, usage, age, years worked and role are listed in the following Table 5. There was a significant positive correlation between age and barriers (r = 0.519, p = < .001). About 27% of the variance was explained by age of participants (r2 = 0.269). There was a significant positive correlation between years worked and barriers (r = 0.451, p = < .001). About 20% of the variance was explained by the number of years worked of participants (r2 = 0.203). There was a significant positive correlation between current professional role and barriers (r = 0.434, p = < .001). About 19% of the variance was explained by current professional role (r2 = 0.188). There was a moderate positive correlation between usage and barriers (r = 0.296, p = < .001). About 9% of the variance was explained by usage (r2 = 0.087). Table 5 Barriers Correlations r p N Age 0.519 <.001 253 Years Worked Within Radiologic Sciences 0.451 <.001 253 Current Professional Role 0.434 <.001 253 Usage Sum 0.296 <.001 253 42 Benefits and Risks Research question three stated, there are no benefits or risks in the implementation and utilization of AI in radiology departments. This question looked to test whether or not there are benefits or risks to AI use in radiology departments in the US based on professional role. Some benefits include increased diagnosis speed and accuracy, assistance with quality control, improved workflow and overall improved patient care. Risks included overreliance on AI software and outside companies, reduced need for technologists, and ethical and legal concerns and potential lack of accountability. The benefits and risks questions were answered using a scale of 1-5 with 1 being “Strongly Disagree”, 2 being “Disagree”, 3 being “Neither Agree nor Disagree”, 4 being “Agree” and 5 being “Strongly Agree”. A one-way ANOVA statistical analysis showed that there are significant findings in benefits and risks based on age of participants, years worked, and imaging role. This finding correlates with the rejection of the null hypothesis that there are neither benefits nor risks for the utilization of AI in radiology. There was a significant difference between age and benefits (F22,230 = 14.251, p = < .001) and about 57% of the variance in benefits was explained by the age of participants (ω2 = 0.565). There was a significant difference between years worked and benefits (F22,230 = 9.793, p = < .001) and about 47% of the variance in benefits was explained by the number of years worked (ω2 = 0.469). There was a significant difference between current role and benefits (F22,230 = 7.336, p = < .001) and about 40% of the variance in benefits was explained by current professional role (ω2 = 0.401). 43 There was a significant difference between age and risks (F18,234 = 25.080, p = < .001) and about 65% of the variance in risks was explained by the age of participants (ω2 = 0.649). There was a significant difference between years worked and risks (F18,234 = 32.672, p = < .001) and about 55% of the variance in risks was explained by the number of years worked (ω2 = 0.550). There was a significant difference between current role and risks (F18,234 = 6.297, p = < .001) and about 48% of the variance in risks was explained by current professional role (ω2 = 0.475). The correlation between benefits, usage, age, years worked, and role are listed in the following Table 6. There was a significant positive correlation between age and benefits (r = 0.697, p = < .001). About 49% of the variance was explained by age of participants (r2 = 0.486). There was a moderate positive correlation between years worked and benefits (r = 0.629, p = < .001). About 40% of the variance was explained by the number of years worked of participants (r2 = 0.396). There was a moderate positive correlation between current professional role and benefits (r = 0.604, p = < .001). About 37% of the variance was explained by current professional role (r2 = 0.365). There was a low positive correlation between usage and benefits (r = 0.526, p = < .001). About 28% of the variance was explained by usage (r2 = 0.277). Table 6 Benefits Correlations Age r p N 0.697 <.001 253 44 Years Worked Within Radiologic Sciences 0.629 <.001 253 Current Professional Role 0.604 <.001 253 Usage Sum 0.526 <.001 253 In addition to benefits, the correlation between risks, usage, age, years worked and role are listed in the following Table 7. There was a significant positive correlation between age and risks (r = 0.674, p = < .001). About 45% of the variance was explained by age of participants (r2 = 0.454). There was a significant positive correlation between years worked and risks (r = 0.587, p = < .001). About 35% of the variance was explained by the number of years worked of participants (r2 = 0.345). There was a significant positive correlation between current professional role and risks (r = 0.596, p = < .001). About 36% of the variance was explained by current professional role (r2 = 0.355). There was a low positive correlation between usage and risks (r = 0.369, p = < .001). About 14% of the variance was explained by usage (r2 = 0.136). Table 7 Risks Correlations r p N Age 0.674 <.001 253 Years Worked Within Radiologic Sciences 0.587 <.001 253 45 Current Professional Role 0.596 <.001 253 Usage Sum 0.369 <.001 253 Evaluation of Findings The purpose of this study was to determine the current usage of AI in radiologic sciences within the U.S. In addition, the perceived barriers to implementing and utilizing AI as well as the potential risks and benefits of AI use were also studied. Using various analysis methods, it was concluded that there are strong correlations between the lack of AI use and the barriers and perceived risks associated with AI usage. Additionally, there was a significant difference in the overall usage and interpretation of benefits based on the age of participants, number of years worked in radiologic sciences and current professional role. In the study, attention was also paid to the facility size and geographic region that participants currently practice to look for trends in current AI usage. Since 1980, when AI technology started advancing till now in the year 2025, it is surprising that across the U.S. there is still a limited amount of AI use in many facilities as shown by this current survey. As stated earlier by the recent literature research, in the United States, there is still a lack of healthcare facilities utilizing AI in radiology and medical imaging (Ungureanu et al., 2025). Thus, healthcare departments are not reaping the benefits of AI including efficiency in reading exams, shorter turnaround times, better understanding of report interpretations in layman terms, quicker access to reports for patients and providers and better workflows in departments to help technologists (Rudolph et al., 2024; Sieber et al., 2024). In order for AI applications to contribute to the 46 improvement of the quality and efficiency of clinical radiology, implementation processes need to be carried out in a structured manner, thereby providing evidence on the clinical added value of AI applications (Strohm, L., Hehakaya, C., Ranschaert, E.R. et al., 2020). Barriers The perceived barriers to AI use that were studied included accuracy and reliability of results from AI sources, high cost of implementation, shortage and unavailability of training to staff and unawareness of AI tools, fear of protecting patient information, disruptions to current workflows and a lack of overall clinical benefits. When looking at the sum of the barriers perceived by the study participants, 45% of the variance in overall usage was attributed to age, 38% was attributed to years worked, and 41% was attributed to professional role. These results show that regardless of age group, the number of years of experience and role within radiology, the perceived barriers surrounding the implementation and use of AI are preventing usage. Risks The perceived risks to using AI that were studied include skill degradation over time with increased use of AI, difficulty in detecting errors produced by inaccurate AI outcomes, potential decrease in job opportunities, and a lack of AI driven accountability for both legal and ethical areas of concern. When evaluating the sum of risk perceived by the participants, 45% of the variance was attributed to age, 35% was attributed to the number of years worked, 36% was attributed to professional role, and about 14% to AI usage. Perceptions of risks certainly can be, in part, attributed to the lack of AI usage, which decreases confidence and understanding of how AI is able to enrich radiology staff 47 (Ungureanu et al., 2025). In similar fashion, Yang et al. (2022) also showed that a common theme among participants in their study included that perceived risks increase as lack of usage and education on AI systems increases. This trend appears to be similar to the findings in this current study and is noted among all participants regardless of age groupings. Benefits The perceived benefits to using AI within radiology that were studied included concepts such as increased diagnosis speed and accuracy, increased quality controls, improved workflow, and improved general patient care and experience. When evaluating the sum of benefits perceived by the participants, about 57% of the variance was explained by the age of participants. About 47% of the variance was explained by the number of years worked, and about 40% of the variance was explained by radiology staff’s current role within their department. These findings can be expected, especially as roles increase from technologists capturing images to radiologists and extenders interpreting results and being more acquainted with various AI systems and how they function. Summary In summary, with the help of one-way ANOVAs, correlations, and descriptive statistics, this research found AI is not being used to its fullest potential. Other findings show that age, years worked, roles within medical imaging, and facility size do play a contributing factor in the usage of AI in medical imaging departments. 48 Chapter 5: Implications, Recommendations, and Conclusions This research was intended to see if technologists and non-technologists (radiologists, managers and physician extenders), working in the radiology field had knowledge of AI currently being used in the U.S. It was the intent to prove there is little knowledge of how AI is being used across the U.S. within many different size facilities, age groups and professional roles. The research highlighted the benefits, risks and barriers that come with the use of AI. Ethical concerns were brought to light with risk and barriers results. Within this chapter we will discuss what was concluded with the results from the survey conducted from 253 responses. Implications Based on the findings of this study, the U.S. has limited use of AI in medical imaging, with barriers related to age, years worked, professional role, and facility size. It is clear that the benefits, barriers to implementation, and risks of AI usage need to be discussed among medical imaging professionals. Through AI education and proper training, medical imaging departments can improve patient care, image quality, resource utilization, and output. (Hussain et al 2025) As the literature was reviewed and the study was completed, it became clear that the U.S. needs to improve its understanding of AI usage in medical imaging. Some areas that AI could be used are information management, quality assurance, diagnoses, and image post-processing but many of the AI/ML applications are still being trained, validated, and tested in research settings before they are utilized in everyday clinical practice (Wichmann et al., 2020). It was also identified through skewed number datasets that the participants of the study are either not 49 using these AI/ML software yet in lieu of the research conducted or are unaware they are using them in everyday practice. The perceived barriers to adapting and implementing AI in radiology modalities consists of but are not limited to, cost, efficacy of AI, and the perceptions of radiologists, imaging technologists, managers, and the general public (Strohm, Lea et al., 2020.) Which also supports the findings of the survey conducted. The barriers identified to utilizing AI in the U.S. include, funding, training, and a resistance to technological changes with a lack of standardization and concerns for patient confidentiality and protection. This correlates to the results rejecting the null hypotheses suggesting that there is limited usage, risks and benefits, and that there are barriers to using and implementing AI in the U.S. One area that would help alleviate many of the barriers and perceived risks would be to better educate radiologic technologists, imaging managers, radiologist extenders, radiologists, physicians and the public about the current usage and multitude of benefits with current AI systems. It has been shown that a key element in successful AI integration is having a leader, or local champion as Strohm, Lea et al. (2020) calls them, to help educate, update, encourage and excite staff about AI and its usage, making it a better experience for both medical staff and public reception alike (Stohm, Lea et al., 2020). The research presented in this paper does highlight the lack of leadership in terms of educating and promoting AI usage and it’s benefits. Recommendations As the literature was reviewed and the study was completed, it became clear that the U.S. needs to improve its understanding and promotion of AI usage in medical imaging. This study has several recommendations that should be considered for future 50 research. These include convenience sampling, self-reporting bias, a lack of longitudinal data, and geographic imbalances, and that it was unclear if the participants of the study knew if they did or did not use AI within their radiologic modality in their day-to-day job roles. Further recommendations may include review of new and emerging findings of published research, and study-based data of currently implemented AI programs throughout the U.S. This study did not include participants from known U.S. AI-based companies whose services are specifically designed for AI implementation in radiological services, and it failed to include known hospitals and clinics who are currently using AI in practices. Future AI usage research as it pertains to barriers, risks, and benefits of implementation in medical imaging may show more positive or negative outcomes if a more focused group were studied. Conclusions In conclusion, the problem researched with the introduction and utilization of AI in healthcare is a lack of understanding of which AI resources are available in the U.S. and the perceived beliefs around AI amongst technologists, radiologist extenders, radiologists and managers. The purpose of this quantitative study is to research the perceptions and challenges related to an under-utilization of AI in healthcare systems. Focusing on the reasons AI is not being used in the U.S were related to funding, training, and a resistance to technological changes with a lack of standardization and concerns for patient confidentiality and protection. Findings suggested that the U.S. has limited use of AI in medical imaging, with known barriers, benefits and risks related to age, years worked, facility size and roles for AI implementation in radiology departments. 51 References Ahn, J. (2004). Electronic portfolios: Blending technology, accountability and assessment. T.H.E. Journal, 31(9), 12-18. Ananny, M., & Crawford, K. (2018). Seeing without knowing: Limitations of the transparency ideal and its application to algorithmic accountability. new media & society, 20(3), 973-989. Bowker, G.C & Star, S.L. (1996) How things (actor-net) work: Classification, magic and the ubiquity of standards. Brady, A. P., Allen, B., Chong, J., Kotter, E., Kottler, N., Mongan, J., ... & Slavotinek, J. (2024). Developing, purchasing, implementing and monitoring AI tools in radiology: practical considerations. A multi-society statement from the ACR, CAR, ESR, RANZCR & RSNA. Canadian Association of Radiologists Journal, 75(2), 226-244. Busch, L. (2011) Standards: recipes for reality. Cambridge: MIT Press. Casey, A., Davidson, E., Poon, M., Dong, H., Duma, D., Grivas, A., ... & Alex, B. (2021). A systematic review of natural language processing applied to radiology reports. BMC medical informatics and decision making, 21(1), 179. Geis, J. R., Brady, A. P., Wu, C. C., Spencer, J., Ranschaert, E., Jaremko, J. L., ... & Kohli, M. (2019). Ethics of artificial intelligence in radiology: summary of the joint European and North American multisociety statement. Radiology, 293(2), 436-440. Hogberg, C. (2004) Stabilizing translucencies: Governing AI transparency by standardization: Big Data and Society. Hong, Y., Lian, J., Xu, L., Min, J., Wang, Y., Freeman, L. J., & Deng, X. (2023). Statistical perspectives on reliability of artificial intelligence systems. Quality Engineering, 35(1), 56-78. Hosny, A., Parmar, C., Quackenbush, J., Schwartz, L. H., & Aerts, H. J. (2018). Artificial intelligence in radiology. Nature Reviews Cancer, 18(8), 500-510. Hussain, M. A., & Warenn, M (2025). AI and radiology: Bridging the gap between peer review, education, and value based care. Applied radiology, 2025(1), 1-3, https://doi.org/10.37549/AR-d-24-0029. 52 Jassar, S., Zhou, Z., Leonard, S., Youssef, A., Probyn, L., Kulasegaram, K., & Adams, S. J. (2025). Educational Competencies for Artificial Intelligence in Radiology: A Scoping Review. Academic Radiology. Jobin, A., Ineca, M. and Vayena, E. (2019). The global landscape of AI ethics guidelines. Nature Machine Intelligence 1(9): 389-399. Khalifa, M., & Albadawy, M. (2024). AI in diagnostic imaging: revolutionising accuracy and efficiency. Computer Methods and programs in biomedicine update, 5, 100146. Klontzas, M. E., Groot Lipman, K. B., Akinci D’Antonoli, T., Andreychenko, A., Cuocolo, R., Dietzel, M., ... & Huisman, M. (2025). ESR Essentials: common performance metrics in AI—practice recommendations by the European Society of Medical Imaging Informatics. European Radiology, 1-13. Kocak, B., Ponsiglione, A., Romeo, V., Ugga, L., Huisman, M., & Cuocolo, R. (2025). Radiology AI and sustainability paradox: environmental, economic, and social dimensions. Insights into Imaging, 16(1), 88. Lobig, F., Subramanian, D., Blankenburg, M. et al. To pay or not to pay for artificial intelligence applications in radiology. npj Digit. Med. 6, 117 (2023). https://doi.org/10.1038/s41746-023-00861-4. Masud R, Al-Rei M, Lokker C. Computer-Aided Detection for Breast Cancer Screening in Clinical Settings: Scoping Review. JMIR Med Inform 2019;7(3):e12660. Mohammad Ihsan Fazal, Muhammed Ebrahim Patel, Jamie Tye, Yuri Gupta. The past, present and future role of artificial intelligence in imaging. European Journal of Radiology. Vol 105 (2018):246-250. ISSN 0720-048X, https://doi.org/10.1016/j.ejrad. 2018.06.020. Nair, A., Ong, W., Lee, A., Leow, N. W., Makmur, A., Ting, Y. H., ... & Hallinan, J. T. P. D. (2025). Enhancing Radiologist Productivity with Artificial Intelligence in Magnetic Resonance Imaging (MRI): A Narrative Review. Diagnostics, 15(9), 1146. Najjar R. Redefining Radiology: A Review of Artificial Intelligence Integration in Medical Imaging. Diagnostics. 2023; 13(17):2760. https://doi.org/10.3390/diagnostics13172760. Nur Ahmed et al. The growing influence of industry in AI research.Science379,884886(2023).DOI:10.1126/science.ade2420. 53 Park, J., Oh, K., Han, K., & Lee, Y. H. (2024). Patient-centered radiology reports with generative artificial intelligence: adding value to radiology reporting. Scientific Reports, 14(1), 13218. Ranschaert, E., Morozov, S., Algra, P. (2019) Artificial Intelligence in Medical Imaging. Springer 2019. Rudolph, J., Johannes Rueckel, Jörg Döpfert, Ling, W. X., Opalka, J., Brem, C., Hesse, N., Ingenerf, M., Koliogiannis, V., Solyanik, O., Hoppe, B. F., Zimmermann, H., Wilhelm Flatz, Forbrig, R., Maximilian Patzig, Boris‐Stephan Rauchmann, Perneczky, R., Peters, O., Priller, J., & Schneider, A. (2024). Artificial intelligence–based rapid brain volumetry substantially improves differential diagnosis in dementia. AlzheimerS & Dementia Diagnosis Assessment & Disease Monitoring, 16(4). https://doi.org/10.1002/dad2.70037. Shank, C.E. (2021). Credibility of soft law from artificial - planning and stakeholder considerations, IEEE Technology and Society Magazine 40(4): 25-36. Shin, Y., Yang, J., Lee, Y. H., & Kim, S. (2021). Artificial intelligence in musculoskeletal ultrasound imaging. Ultrasonography, 40(1), 30-44. Strohm, L., Hehakaya, C., Ranschaert, E.R. et al. Implementation of artificial intelligence (AI) applications in radiology: hindering and facilitating factors. Eur Radiol 30, 5525– 5532 (2020). https://doi.org/10.1007/s00330-020-06946-y. Sieber, V., Rusche, T., Yang, S., Bram Stieltjes, Fischer, U., Stefano Trebeschi, Cattin, P., Nguyen-Kim, D. L., Marios-Nikos Psychogios, Lieb, J. M., & Sporns, P. B. (2024). Automated assessment of brain MRIs in multiple sclerosis patients significantly reduces reading time. Neuroradiology, 66(12), 2171–2176. https://doi.org/10.1007/s00234-02403497-7 Ungureanu, A.-M., Matei, S.-C., & Malita, D. (2025). Controversies in the Application of AI in Radiology—Is There Medico-Legal Support? Aspects from Romanian Practice. Diagnostics, 15(2), 230–230. https://doi.org/10.3390/diagnostics15020230. U.S. Food & Drug Administration. (2025). Artificial Intelligence and Machine Learning (AI/ML)-Enabled Medical Devices. FDA. https://www.fda.gov/medicaldevices/software-medical-device-samd/artificial-intelligence-and-machine-learning-aimlenabled-medical-devices?utm_medium=email&utm_source=govdelivery. U.S. Government Printing Office. (2006). Catalog of U.S. Government publications: New electronic titles. Vimalesvaran, K., Robert, D., Kumar, S., Kumar, A., Narbone, M., Dharmadhikari, R., ... & Lowe, D. J. (2024). Assessing the effectiveness of artificial intelligence (AI) in 54 prioritising CT head interpretation: study protocol for a stepped-wedge cluster randomised trial (ACCEPT-AI). BMJ open, 14(6), e078227. Wenderott, K., Krups, J., Luetkens, J. A., Gambashidze, N., & Weigl, M. (2024). Prospective effects of an artificial intelligence-based computer-aided detection system for prostate imaging on routine workflow and radiologists’ outcomes. European Journal of Radiology, 170, 111252. Wichmann, J. L., Willemink, M. J., & De Cecco, C. N. (2020). Artificial intelligence and machine learning in radiology: current state and considerations for routine clinical implementation. Investigative Radiology, 55(9), 619-627. Winslade, J., & Monk, G. (2001). Narrative mediation: A new approach to conflict resolution. San Francisco, CA: Jossey-Bass. Yang, Ene, I.C., Arabi Belaghi, R. et al. Stakeholders’ perspectives on the future of artificial intelligence in radiology: a scoping review. Eur Radiol 32, 1477–1495 (2022). https://doi.org/10.1007/s00330-021-08214-z. Yeasmin, M. N., Al Amin, M., Joti, T. J., Aung, Z., & Azim, M. A. (2024). Advances of AI in image-based computer-aided diagnosis: A review. Array, 23, 100357. Zanardo, M., & Sconfienza, L. M. (2025). Translational approaches for radiation protection: treating patients, not their images. European Radiology, 1-3. Zech, J. R., Ezuma, C. O., Patel, S., Edwards, C. R., Posner, R., Hannon, E., Williams, F., Lala, S. V., Ahmad, Z. Y., Moy, M. P., & Wong, T. T. (2024). Artificial intelligence improves resident detection of pediatric and young adult upper extremity fractures. Skeletal Radiology,66(5). https://doi.org/10.1007 /s00256-024-04698-0. Zuo, Y. (2025). Exploring the synergy: AI Enhancing Blockchain, Blockchain Empowering AI, and Their Convergence Across IoT Applications and Beyond. IEEE Internet of Things Journal. (Volume:12, Issue 6, 15 March 2025). 55 Appendices 56 Appendix A: IRB Approval Letter 57 Appendix B: Survey Artificial Intelligence within Radiologic Sciences Q1 You are invited to participate in a research study exploring the use of Artificial Intelligence (AI) among technologists and practitioners within the Radiologic Sciences. You were selected as a possible subject, because you are currently employed within the United States, within a profession associated with Radiologic Sciences. We ask that you read this form and ask any questions you may have before agreeing to be in the study. The study is being conducted by Tanya Nolan, Adam Fisher, Codi Harbour, Katlyn Kiger, Mike Oveson, Kim Poss, and Emilee Scott as part of an MSRS research project at Weber State University. STUDY PURPOSE The purpose of this study is to research the use of AI among Radiologic Sciences professionals within the United States. As there is limited research within the country regarding practical applications of AI within the Radiologic Sciences, research questions highlight the current opportunities, resources, challenges, and benefits perceived by professionals within the Radiologic Sciences. NUMBER OF PEOPLE TAKING PART IN THE STUDY: If you agree to participate, you will be one of a minimum of 50 subjects who will be participating in this research. PROCEDURES FOR THE STUDY: If you agree to be in the study, you will be asked to complete a 15-minute electronic survey including questions on use and perceptions of AI within the Radiologic Sciences and some demographics. RISKS OF TAKING PART IN THE STUDY: The risk of completing the survey may be discomfort in answering questions regarding personal perceptions. BENEFITS OF TAKING PART IN THE STUDY: You will not receive payment for taking part in this study. ALTERNATIVES TO TAKING PART IN THE STUDY: There are no alternatives to taking part in the study. CONFIDENTIALITY Efforts will be made to keep your personal information confidential. We cannot guarantee absolute confidentiality. Your personal information may be disclosed if required by law. Your identity will be held in confidence in reports in which the study may be published. No personal identifiers will be included within the survey, and all data is kept in a password-protected system, available solely to investigators for educational purposes. All data will be destroyed within 2 years. Organizations that may inspect and/or copy your research records for quality assurance and data analysis include groups such as the study investigator and his/her research associates, the Weber State University Institutional Review Board or its designees, the study sponsor, and (as allowed by law) state or federal agencies, specifically the Office for Human Research Protections (OHRP) and the Food and Drug Administration (FDA) [for FDA-regulated research and research involving positron-emission scanning], the National Cancer Institute (NCI) [for research funded or supported by NCI], the National Institutes of Health (NIH) [for research funded or supported by NIH], etc., who may need 58 to access your medical and/or research records. CONTACTS FOR QUESTIONS OR PROBLEMS For questions about the study, contact the researcher Tanya Nolan at 801626-8172. For questions about your rights as a research participant or to discuss problems, complaints or concerns about a research study, or to obtain information, or offer input, contact the Chair of the IRB Committee IRB@weber.edu. VOLUNTARY NATURE OF STUDY Taking part in this study is voluntary. You may choose not to take part or may leave the study at any time. Leaving the study will not result in any penalty or loss of benefits to which you are entitled. Your decision whether or not to participate in this study will not affect your current or future relations with your clinical affiliate or the University. SUBJECT’S CONSENT By selecting “yes” below, you have consented to participate in the study. o Yes (1) o No (2) Q2 Please indicate your level of Artificial Intelligence (AI) usage in the following areas of your professional practice on a scale from 0 (Never) to 10 (Always). Never Rarely Sometimes Often Always 0 Information Management (e.g., workflow automation, patient scheduling, report generation) () Diagnoses (e.g., lesion detection, image interpretation, clinical decision support) () Post-Image Processing (e.g., segmentation, 3D reconstruction, image enhancement) () Quality Assurance (e.g., error detection, protocol compliance, peer review support) () 1 2 3 4 5 6 7 8 9 10 59 Q3 Please rate the extent to which you agree with the following statements as barriers to adopting artificial intelligence (AI) into your professional practice. Strongly Neither Disagree Strongly Disagree agree nor Agree (4) (2) Agree (5) (1) disagree (3) I am concerned about the accuracy and reliability of AI results (1) I feel the cost of implementing AI is too high. (2) I worry about protecting patient information. (3) I believe my team lacks training or familiarity with AI tools. (4) I do not believe AI has a clear clinical benefit and may disrupt workflow. (5) o o o o o o o o o o o o o o o o o o o o o o o o o 60 Q4 Please indicate your level of agreement with the following statements about the benefits of AI in radiology, based on your professional role. Neither Strongly Disagree agree nor Strongly Disagree Agree (4) (2) disagree agree (5) (1) (3) AI improves diagnostic accuracy. (1) AI speeds up image interpretation and reporting. (2) AI assists in quality control and protocol adherence. (3) AI offers valuable support for less experienced staff. (4) AI enhances workflow efficiency across departments. (5) AI improves patient care and communication. (6) o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o 61 Q5 Please indicate your level of agreement with the following statements about the risks of AI in radiology, based on your professional role. Neither Strongly Disagree agree nor Strongly Disagree Agree (4) (2) disagree agree (5) (1) (3) Overreliance on AI may lead to skill degradation among radiologists or technologists. (1) AI tools may produce errors that are difficult to detect. (2) AI implementation may reduce job opportunities or shift responsibilities. (3) Ethical or legal concerns limit the safe use of AI in practice. (4) AI increases dependence on external vendors and tech companies. (5) There is a lack of clear accountability when AI makes incorrect suggestions. (6) o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o 62 Q6 What is your age? o 18-27 (1) o 28-37 (2) o 38-47 (3) o 48-57 (4) o 58 + (5) o Prefer not to answer (6) Q7 How many years have you worked within the Radiologic Sciences? o 5 years and under (1) o 6-10 years (2) o 11- 15 years (3) o 16-20 years (4) o 21-25 years (5) o 25 + years (6) Q8 Please identify your current professional role within the Radiologic Sciences. o Technologist (1) o Imaging Manager (2) o Radiologist Extender (RA, RPA, PA) (3) o Radiologist / Physician (4) 63 Q9 In which specialties do you have a responsibility? (choose all that apply) ▢ ▢ ▢ ▢ ▢ ▢ ▢ ▢ ▢ ▢ ▢ Bone Densitometry (1) Cardiac Interventional (2) Computed Tomography (3) Magnetic Resonance (4) Mammography (5) Medical Dosimetry (6) Nuclear Medicine (7) Radiation Therapy (8) Radiography (9) Sonography (10) Vascular Interventional (11) 64 Q10 How would you describe the location of your employment? o Armed forces (1) o Outpatient Facility or Imaging Center (2) o Rural Hospital 1-25 beds (3) o Small Hospital 25-100 beds (4) o Medium Hospital 100-499 beds (5) o Large Hospital >500 beds (6) o Mobile Unit (7) o Physician's Office or Clinic (8) o Home Health (9) o Other (10) Q11 Please type the full name of the state within the U.S. in which you work. ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ Adam Fisher, Codi Harbour, Katlyn Kiger, Tanya Nolan, Michael Oveson, Kimberly Poss, Emilee Scott (2025). Research Project Survey. Retrieved from https://weber.co1.qualtrics.com/jfe/form/SV_aXBnBFFwFMLnm74 65 Appendix C: Examples of Participant Invitations A QR code and link to the Qualatrics survey were shared with all group members to distribute to their imaging colleagues and contacts. Surveys were distributed via a variety of platforms including, LinkedIn, Sonography Underground, Facebook, Instagram, work emails, and personal emails. The link and QR code were distributed and shared for over two weeks to collect data for this research study. 66 67 Appendix D: Tables Table 8 Facility Size Descriptives N Percentage Outpatient Facility or Imaging Center 17 10.5% Rural Hospital 1-25 beds 9 5.6% Small Hospital 25-100 beds 7 4.3% Medium Hospital 100-499 beds 78 48.1% Large Hospital >500 40 24.7% Physician’s Office or Clinic 4 2.5% Other 7 4.3% Total 162 100 68 Table 9 Current Professional Role Descriptives N Percentage Technologist 133 82.6% Imaging Manager 10 6.2% Radiologist Extender (RA, RPA, PA) 5 3.1% Radiologist/Physician 13 8.1% Total 161 100 69 Table 10 Years Worked Descriptives N Valid Percentage 5 years and under 51 31.5% 6-10 years 32 19.8% 11-15 years 22 13.6% 16-20 years 18 11.1% 21-25 years 12 7.4% 25+ years 27 16.6% Total 162 100% 70 Table 11 Age Descriptives N Valid Percentage 18-27 34 31.5% 28-37 46 19.8% 38-47 36 13.6% 48-57 34 11.1% 58+ 9 7.4% Prefer not to answer 2 16.6% Total 162 100 71 Appendix E: Figures Figure # 4 Number of Years Worked Figure # 5 Professional Role 72 Figure # 6 Age of Participants