1 Introduction

Artificial intelligence (AI) is an umbrella term that incorporates various algorithms including machine learning, deep learning, and natural language processing (Potočnik et al., 2023). A recent report by McKinsey & Company (McKinsey & Company, 2023, p. 2) describes AI in the following way:

AI is a machine’s ability to perform the cognitive functions we associate with human minds, such as perceiving, reasoning, learning, interacting with an environment, problem-solving, and even exercising creativity… [T]he value of artificial intelligence isn’t in the systems themselves but in how companies use those systems to assist humans—and their ability to explain to shareholders and the public what those systems do—in a way that builds and earns trust.

In recent years, AI has emerged as a transformative technology in the field of medical care, in radiology in general, and in medical imaging in particular. These advancements have changed the way that healthcare providers diagnose and treat various conditions.

In this paper, we describe the evolution of AI in medical imaging technology. Motivating our focus on medical imaging technology are the myriad applications of the technology, all of which have the potential for benefiting society—through early disease detection, enhanced diagnostic accuracy, and efficient workflow dynamics— and the pervasive evidence that the global private sector perceives the need for these applications. For example, it was recently estimated that the global private sector will invest over $6 billion in U.S. dollars in AI in 2022; this investment amount surpasses the global private sector’s investment in any other technology.Footnote 1,Footnote 2

The remainder of this paper is organized as follows. In Section II, we overview the history of the adoption of AI technology in medical imaging. This overview provides context for the remaining sections of the paper. In Section III, we explore the importance of AI in the medical imaging field, highlighting its impact on practices, diagnoses, and patient care. This historical emphasis provides a segue to our discussion of the economic impacts of the technology in Section IV. In that section, we focus broadly on the economic impact of AI in medical imaging. We analyze the cost-efficiency, labor outcomes, and broader economic implications offering insight into the opportunities and challenges of AI adoption in the medical imaging field. The paper concludes in Section V with reflections on the transformative journey of AI in medical imaging emphasizing its implications for healthcare, social benefits, and the collaborative potential of AI and medical imaging in shaping a sustainable and impactful future.

2 History of AI related to medical imaging

The field of radiology was an early adopter of a breakthrough technology, which occurred as a result of the remarkable discovery of X-rays (Gore, 2020).Footnote 3 In the late 19th century, Wilhelm Conrad Roentgen’s discovery of X-rays marked the dawn of medical imaging (Pakdemirli, 2019). X-ray technology provided physicians and diagnosticians with the unprecedented ability to peer inside the human body without invasive procedures thus setting the stage for a transformation in healthcare.

X-ray technology continued to evolve during the 20th century, delivering progressively clearer images of the body. Radiological departments became integral parts of hospitals, driving growth in the healthcare sector by presenting unprecedented opportunities to enhance quality and efficiency of patient care (Najjar, 2023). However, challenges to the broad adoption of X-ray technology remained. For example, traditional X-ray images relied on film, requiring time-consuming chemical development processes. Such hurdles, along with the need for the training of technicians in procedures and physicians in how to read and interpret film, prompted investments, especially private-sector investments in research and development (R&D) to improve imaging technologies (Mayo et al., 2020).

The development of ultrasound technology began in the 1940s; however, pioneering advancements in ultrasound application in medical applications, including medical imaging, were not reported until the 1950s (Lanza, 2020).Footnote 4 The term “Artificial Intelligence” was not introduced till 1956 by John McCarthy and his team at Dartmouth College (Shen et al., 2021), and early ultrasound technology did not initially incorporate AI algorithms for image interpretation. While the applications of ultrasound during the 1940s were experimental, these early efforts at ultrasound imaging represented an important step in the introduction and advancement of AI for medical imaging applications (Kendall et al., 2007). Over time, AI has been increasingly applied to ultrasound, utilizing machine learning techniques to improve the accuracy and efficiency of image analysis and interpretation.

Efforts to integrate the use of AI in medical imaging occurred in the 1960s, brought about through Computer-Aided Diagnosis (CADx).Footnote 5 CADx was used to improve chest x-ray and mammography procedures (Pesapane et al., 2018). Computed Tomography (CT) imaging procedures advanced during the 1970s.Footnote 6 CT scans used multiple X-ray beams to create cross-sectional images which improved diagnostic accuracy. The CADx breakthrough technology had significant health-related implications, and thus this technology rapidly diffused throughout hospitals and medical facilities (Hillman, 1986).

During the 1980s, Magnetic Resonance Imaging (MRI) was introduced into the healthcare market. It improved medical imaging by using magnets and radio waves to create detailed images of soft tissues thus avoiding the need for harmful radiation (Gore, 2020).Footnote 7 This advancement not only prioritized safety but also opened new markets for manufacturers, healthcare providers, and investors. MRI quickly became a tool in the fields of neurology and musculoskeletal diagnostics, as well as radiology.

By the late 20th century, applications of digital technology began to challenge the continued use of analog technology in medical imaging.Footnote 8 Radiography, fluoroscopy, and mammography embraced this transition by offering image acquisition and efficient storage and transmission methods (Rajpurkar & Lungren, 2023). The advent of Picture Archiving and Communication Systems (PACS) streamlined healthcare operations while bringing benefits through reduced costs associated with film-based processes (Panayides et al., 2020).Footnote 9

A new era for the application of AI in medical imaging began in the early 2000s. Machine learning techniques and learning algorithms revolutionized image interpretation by providing exceptional accuracy in detecting and classifying abnormalities (Agboklu et al., 2024). Radiologists began leveraging AI as a tool to enhance their capabilities and optimize workflow efficiency. In 2017, 36 million MRI scans were performed in the United States (Panayides et al., 2020). Convolutional Neural Networks (CNNs) gained popularity for their ability to focus on image classification and detection tasks, and as such they were used in detecting breast cancer from mammograms and identifying abnormalities in radiological scans.Footnote 10

Generative AI was introduced in 2018 (Lorenz et al., 2023) with the emergence of Generative Adversarial Networks (GANs).Footnote 11 GANs led to applications in image generation, text generation, and natural language processing, creating contents and chatbots. In medical imaging, these AI techniques are currently being used to enhance image quality, generate synthetic medical images for training purposes, and aid in the development of advanced diagnostic tools. By 2020, the application of AI to medical imaging shifted towards more complex tasks, including image segmentation, organ localization Footnote 12, and disease diagnosis (Shamshad et al., 2023). Models like the Vision Transformer (ViT) gained traction for its ability to capture long-range dependencies in medical images within the body.Footnote 13 This innovation allowed for the analysis of medical images where features or patterns required interactions across relatively distant or spatially separated areas within the image.

Today, medical imaging technology continues to evolve. Applications of AI have expanded beyond traditional radiology, aiding in pathology, dermatology, and even genomics. Imaging techniques—3D printing, molecular imaging, and functional MRIFootnote 14—are pushing the boundaries of diagnostics and treatment planning (Agboklu, 2024).

Looking toward the future, the combination of AI and medical imaging holds promise for improved precision and accuracy in medicine. AI algorithms will be able to analyze genetic information to tailor treatments on an individual basis (Sollini et al., 2020). The role of AI in disease detection and predicting disease progression will have an impact on healthcare costs as well as overall outcomes (Ahmed et al., 2020).

To summarize, the history of medical imaging technology is marked by innovation and advancement fueled by a commitment to enhance healthcare. With the progress of technology, society is provided with advanced tools and accurately tailored methods for diagnosing and treating medical ailments, ultimately contributing to the overall betterment of individuals (Harris, 2023). This shift from a one size fits all approach to personalized medicine not only improves patient outcomes but also enhances AI’s role in early detection and predicting disease progression, thus optimizing the allocation of healthcare resources. By forecasting on how these diseases may evolve, healthcare providers can more efficiently allocate scarce resources thereby resulting in cost effectiveness for both producers and consumers of healthcare (Lartey et al.,2023).

Figure 1 provides a visual summary journey, following the discussion above, through the transformative impact of AI on healthcare, especially medical imaging.

Fig. 1
figure 1

Source: Prepared by the authors

Evolution of AI in Medical Imaging.

Full size image

3 Importance of AI to the medical imaging field

Imaging technologies are essential for diagnosing abnormalities and guiding therapy. They involve advanced techniques for creating visual representations that provide medical professionals with valuable insight into their patients’ conditions, thus significantly improving healthcare professionals’ understanding of the patients’ medical condition (Flower, 2012; National Research Council, 1996). Serving essential functions in assessing, treating, and preventing illnesses, medical imaging plays an important role in diagnosing a broad spectrum of medical abnormalities and diseases. These abnormalities range from traumatic injuries to cancer and cardiovascular diseases. And skilled professionals, including oncologists and internists, rely on medical imagining for an accurate diagnosis and an effective treatment roadmap (Laal, 2013).

While medical imaging undeniably enhances the precision and accuracy of disease diagnosis leading to more effective treatments, it is important to acknowledge that reliance on medical imaging has certain drawbacks including elevated expenses and potential patient harm such as incidental findings, overdiagnosis, heightened anxiety among patients, and the increased risk of cancer due to radiation exposure.

Studies have suggested that a substantial proportion of imaging procedures, possibly exceeding 30%, lack necessity resulting in a considerable financial burden of approximately $30 billion annually in the United States (McGinnis et al., 2013). But despite these negative social impacts, between 2000 and 2006 the United States saw a rise in the use of CT, MRI, and ultrasound as a diagnostic tool (Smith-Bindman et al., 2019). A more recent study found that from 2000 to 2016, CT and MRI use increased among adults in seven U.S. healthcare systems and in the province of Ontario, Canada, but the rate of increase has slowed in more recent years. For children, imaging rates rose, except for CT usage which stabilized and decreased more recently (Smith-Bindman et al., 2019).

Social implications associated with medical imaging in general are significant, particularly with the public sector’s continual use of AI to enhance the application of this technology. For example, the National Institutes of Health (NIH) has established programs to:Footnote 15

… propel biomedical research forward by setting the stage for widespread adoption of artificial intelligence (AI) that tackles complex biomedical challenges beyond human intuition.

And the National Institute of Biomedical Imaging and Bioengineering (NIBIB) within the NIH is funding research to, among other things, advance the early diagnosis of Alzheimer’s disease (AD) through an analysis of brain networks:Footnote 16

AD-related neurological degeneration begins long before the appearance of clinical symptoms. Information provided by functional MRI (fMRI) neuroimaging data, which can detect changes in brain tissue during the early phases of AD, holds potential for early detection and treatment. The researchers are combining the ability of fMRI to detect subtle brain changes with the ability of machine learning to analyze multiple brain changes over time. This approach aims to improve early detection of AD, as well as other neurological disorders including schizophrenia, autism, and multiple sclerosis.

While having chronicled the applications and significance of advances in AI in the field of medical imaging to date, it is equally important to explore the often-underestimated aspect of the direct, indirect, financial, and social costs associated with integrating AI into this specialized domain. Understanding these financial considerations is vital as they play a pivotal role in the successful implementation of sustainable AI-driven solutions in healthcare.

4 Economic impacts of AI in medical imaging

The opportunity cost of not fully embracing AI extends beyond immediate inefficiencies such as lower standards in products and services in the health sector; it also includes the unrealized benefits and advancements that could have propelled us into a more intelligent and competitive technological landscape (Pagallo et al., 2023).

As the technology of AI evolves and thus changes how it is used in medical imaging, policymakers are faced with the challenge of establishing guidelines to ensure the effective integration and equitable use of this technology. The interactions among key stakeholders, including both humans (such as physicians and patients) and non-humans (such as AI technologies) in the healthcare sector, may foster more robust hybrid knowledgeFootnote 17. This dynamic could contribute to a more sustainable introduction of AI in the health sector (Kannelønning, 2024). This growth in the application of AI in general has, however, been met with mixed sentiments in the workforce. Early on, at least one eminent computer scientist declared (Lynch, 2017):Footnote 18

AI, like electricity or the internet, is a general-purpose technology that will have a profound impact on our society, spurring economic growth, creating new opportunities, and reshaping the way we live and work.

Many workers are uneasy about AI. They either have reservations of its role, or they remain unconvinced of its value in the broader workplace (Tucci, 2024). Without worker trust, the benefits of AI may not be fully realized.Footnote 19 This angst is not unexpected. David Wells, chief financial officer at Netflix, noted when asked how humans and AI might evolve together (Anderson & Raine, 2018):Footnote 20

Technology progression and advancement has always been met with fear and anxiety, giving way to tremendous gains for humankind as we learn to enhance the best of the changes and adapt and alter the worst.

Such concerns affect the healthcare sector, and perhaps those who are involved in medical imaging. Early on, McKinsey & Company (2017, p. 23) suggested that AI will affect the supply chain in health care in four ways, all of which will reduce health care costs and increase patient health:

  • Predict disease, identify high-risk patient groups, and launch prevention therapies.

  • Automate and optimize hospital operations; automate diagnostic tests and make them faster and more accurate.

  • Predict cost more accurately, focus on patients’ risk reduction.

  • Adapt therapies and drug formulations to patients, use virtual agents to help patients navigate their hospital journey.

More recently, McKinsey & Company (2023, pp. 58–62) assessed qualitatively the economic benefits associated with AI in healthcare in the following way:

  • AI can identify public-health threats and the most at-risk patients.

  • AI can help medical professionals diagnose disease and improve operations.

  • Insurers can devise new ways to encourage preventive care and incentivize providers.

  • Doctors will be able to tailor treatments—even drugs—to individual patients.

  • Virtual agents can serve as primary touchpoints for patients.

Regarding AI in diagnosis, the GAO (2022, p. 1) opined the following point of view:

Artificial intelligence (AI) has emerged as a powerful tool for solving complex problems in diverse domains. Machine learning (ML), a subfield of AI, could revolutionize diagnosis by augmenting clinical diagnostics practice resulting in earlier and better diagnoses, lives saved, and avoided costs of time and money. In recent years, for example, ML technology was reported to be equivalent to medical professionals in interpreting medical data from fields like radiology and dermatology. ML technology can assist medical professionals in completing repetitive tasks without getting tired and flagging potential medical issues at the point of care.

Unfortunately, at this current point in time, there are not sufficient data to quantify the economic benefits, or the economic costs associated with improved diagnostic procedures, in general, or improved medical imaging technology applications, in particular.

These benefits include enhancing the drug development process by improving efficiency and reducing time and costs, detecting diseases earlier, and analyzing more accurately medical data (GAO, 2022). Yet, the surge in the adoption of AI technologies in healthcare has raised several concerns and questions among researchers and healthcare professionals (Harris, 2023). As noted above, concerns revolve around accuracy, security, and privacy (Mökander et al., 2022), as well as ensuring the reliability and protection of medical data. As AI systems process vast amounts of data—encompassing personal, private, and sensitive information of patients—there is a heightened risk if these data fall into the wrong hands. The likelihood of such happening raises privacy concerns, thus requiring detailed attention to the examination of each image to identify and remove potential identifiable information (Prevedello et al., 2019).

A recent survey by Alexander et al. (2020) found that one of the biggest barriers to the adoption of AI in medical imaging by radiologists was their concern about AI’s diagnostic capabilities in more complex patients and diseases. These concerns highlight potential risks, including misdiagnosis and perpetuation of biases in AI systems. Generative AI also raises intellectual property concerns, such as unlicensed content in training data, copyright, patent, trademark issues with AI creations, and ownership of AI-generated work (Lorenz et al., 2023). Given these concerns, some have argued that regulations are needed regarding the ethical use of AI in medical imaging, patient care, and workforce management to safeguard patients and healthcare professionals alike (Mudgal & Das, 2020).

As the use of AI technology expands, international regulation discussions grow. The EU’s proposed Artificial Intelligence Act (AIA) outlines recommendations for a wide range of AI applications. The proposed regulations introduce varying rules based on risk levels associated with different AI systems. These recommendations include guidelines for ethical AI development, usage, and accountability. The AIA emphasizes the importance of ensuring transparency, fairness, and human oversight in AI systems. Additionally, it addresses potential risks and concerns, such as bias mitigation, data privacy, and the prevention of undue concentration of power.

The United States also introduced the Algorithmic Accountability Act of 2022, but it has not broadly regulated AI yet. This Act, proposed in both H.R. 6850 and S. 3572 during the 117th Congress, reflects a growing awareness of the need for regulatory frameworks in the rapidly evolving nature of artificial intelligence. The Act aims to address the potential risks associated with the deployment of AI, including issues related to transparency, fairness, and accountability (Harris, 2023).

The AIA is currently under negotiations in the EU, and the final text of the legislation may shape future EU-U.S. alignment or divergence in AI (Mökander et al., 2022). The active development and evaluation of AI technologies in medical imaging, including the need for ethical regulations as a result of their potential risks, highlight the ongoing evolution and resulting application of AI in the healthcare sector. Thus far, AI has had a positive impact on job quality, enhancing the overall quality of work through a general increase in workers’ well-being. However, the potential negative effects on job quality may take time to manifest themselves (OECD, 2023). The integration of AI into medical imaging will inevitably shape the healthcare labor workforce, prompting the need for proactive measures to address workforce dynamics (World Health Organization, 2021). Along these lines, Muro et al. (2019) projected a number of years ago that within the next decade, approximately thirty-six million jobs spanning multiple industries are at significant risk of being highly impacted by automation. Such profound shift in the workforce necessitates a strategic approach to ensure that the workforce adapts and is capable of thriving in this AI-enhanced environment. Along this line, Hazarika (2020, p. 244) has pointed out:

AI has a huge potential to alleviate some of the challenges that healthcare providers face. Optimizing the benefits of AI will require a balanced approach that enhances accountability and transparency while facilitating innovation, fostering responsible access to data to further develop computing abilities and building trust between providers, patients, researchers, and innovators. Looking into the future—taking cues from the history of automation—AI is unlikely to displace humans, but will definitely redefine their roles and establish itself as an indispensable cognitive assistant.

As a response to these changes in the workforce, the 118th Congress of the United States introduced a total of ninety-four separate bills, none of which has yet been approved. These bills covered a wide array of topics, one of which was the importance of AI training for federal employees (Harris, 2023). The focus on AI training in federal agencies reflects the recognition that building employee trust in AI systems is essential for their effective integration into government operations and by extension in the healthcare, especially medical imaging, field.

5 Conclusions

The evolution of medical imaging technology, from the initial discovery of X-rays to the integration of advanced AI systems, has marked a transformative journey in healthcare in general and in medical imaging in particular. This journey reflects not only technological advancement but also the commitment to enhance patient outcomes and the overall well-being of individuals. While AI promises significant advancements, concerns about accuracy, security, and privacy must be addressed. The ethical use of AI highlights the need for thoughtful regulations to address these concerns.

The proposed Artificial Intelligence Act in the EU and the Algorithmic Accountability Act of 2022 in the United States illustrate the global recognition of the need for regulatory frameworks in an ever-evolving field of artificial intelligence. These regulations aim to ensure transparency, fairness, and accountability in AI systems, mitigating potential risks and fostering innovation.

Moreover, the economic impacts of AI in medical imaging extend beyond technological advancements, shaping workforce dynamics and job quality. As AI becomes an indispensable cognitive assistant, proactive measures, such as AI training for federal employees, are essential to ensure a workforce that adapts and thrives in this AI-enhanced environment.

Looking ahead, the collaboration between AI and medical imaging holds the promise of personalized medicine, data-driven treatments, and optimized healthcare resource allocation. By understanding the financial considerations and potential challenges associated with AI integration, policymakers and healthcare professionals can collaboratively shape a sustainable and impactful future for AI in medical imaging.

In essence, the transformative impact of AI in medical imaging extends beyond technological milestones, resonating with the core principles of enhancing healthcare, fostering innovation, and ensuring ethical practices. The merging of AI and medical imaging emerges not just as a technological advancement but as a pivotal force in shaping the future of healthcare.

Given the growth of AI in medical imaging and its importance in transforming healthcare, it would be prudent for future research endeavors to prioritize understanding the diffusion dynamics of AI medical imaging technologies over time. Longitudinal studies coupled with advanced data analytics methodologies could provide comprehensive insights into the adoption patterns and market distribution. By tracking technology adoption rates and utilization trends across healthcare institutions and research organizations, researchers can elucidate the evolution of AI medical imaging technology within the broader healthcare landscape. Leveraging diverse methodologies, researchers can address the challenges associated with defining market percentages and gain a holistic understanding of technology adoption patterns within the evolving landscape of medical imaging.

In conclusion, while this study’s focus, as illustrated in Fig. 1, centers on AI within medical imaging, it is important to note that the evolution and adoption rates of AI technologies vary significantly across different sectors and industries. Medical imaging presents unique challenges and opportunities for AI integration, given its reliance on intricate data interpretation, stringent diagnostic accuracy standards, and the necessity for seamless integration into clinical workflows.

The limitations for generalization to other technologies arise from the distinctive considerations inherent in medical imaging, including the interpretability of AI algorithms, regulatory approval processes, and ethical implications related to patient data privacy and security. Therefore, while this study provides valuable insights into AI’s impact on medical imaging applications, it is important to warrant caution when extending these findings to other technological domains.