How AI is driving advances in medical image management

Studies show that radiologists suffer from burnout more than most other medical professionals, in part due to long hours, complex cases and inefficiencies in radiology workflows. And the pressures facing this group of professionals are expected to worsen in the next decade. This may unfortunately be contributing to a shortage in the field, and by 2033, there could be a shortfall of nearly 42,000 medical imaging professionals in the United States.

Recently, advanced capabilities such as artificial intelligence (AI) and machine learning have shown great potential for alleviating the overwhelming workloads of radiologists, enabling them to work smarter — not harder. AI assists in making workflows more efficient, and enables radiologists to gather greater insights from the vast amounts of data being generated by imaging solutions. This, in turn, can result in improved patient outcomes and reduced costs of care.

Looking ahead, AI can be leveraged at an even greater capacity to drive many more advances that will make imaging more valuable across the entire healthcare continuum. 

The Promise of AI in Medical Image Management

Of all the potential uses of AI in healthcare, medical imaging holds the most promise, namely because of the sheer amount of data it produces. Hospitals generate about 50 petabytes of data every year, and about 90% is from medical imaging technology such as magnetic resonance imaging (MRI), computed tomography (CT) and positron emission tomography (PET) scans, ultrasound and other tools. Yet, it is estimated that 97% of that imaging data goes untapped.

Advanced analysis, including AI, is already being applied in imaging applications to address this challenge. For example, AI has enabled major innovation in stroke management. By embedding AI into CT scans to help quickly detect brain hemorrhages, radiologists and emergency department physicians can confer and determine whether treatment must be done right away at that hospital, or if the patient can be sent to a facility that specializes in stroke care. In this application, AI combined with CT angiography enables faster diagnosis, enhancing provider efficiency and decision-making capabilities, and resulting in better care and potentially better outcomes for patients.

Overcoming AI Adoption Challenges

While there is great promise for AI in medical image management, the use of AI in dedicated point solutions also creates challenges to its adoption. Because the technology is frequently used to address only one clinical question — such as whether a patient is suffering from a brain bleed, using a single modality, like CT angiography — it can be difficult to justify the cost and overall implementation of an AI-based solution.

Notably, recent studies have shown that only 56% of radiologists currently use some sort of AI, and less than 38% reported exposure to any of the five common AI use cases. These include the study identified as “tagging images to ensure those for critical patients are reviewed first; optimizing workflow for overall productivity; automating part of the image analysis; providing clinicians with decision support; and enhancing imaging quality.” In fact, only 10% have regarded themselves as “very familiar” with these applications.

Healthcare also lags behind other sectors in adoption of AI because it operates in such a highly regulated environment. Only in the last few years have AI-based solutions been approved for clinical use.

With so many new offerings, it can be difficult to evaluate the whole market to determine which solution(s) will work best for a health system’s modalities, radiologists, referrers and ultimately their patients. And many organizations have yet to take a holistic look at AI solutions, develop an overall business case that financially justifies use of AI-based clinical applications, and create a roadmap to integrate it into their workflows.

Reaching the Quadruple Aim in Healthcare With AI

Today, however, the dynamics are slowly shifting — and for good reason. There is now a clearer focus on reimbursement for investments in AI and machine learning, from New Technology Add-On Payment (NTAP) and CPT codes, to Medicare Coverage of Innovative Technology (MCIT). A more defined path to reimbursement and an increasing understanding of the value of AI-based clinical applications will no doubt spark more implementations across all aspects of medical imaging. And in the process, it will help health systems realize the promises of AI. This includes making workflows more efficient, reducing the stress experienced by overworked radiologists and speeding interpretation of the images so patients have the chance for better outcomes.

But we expect to see even greater benefits as some trailblazing health systems prove out broader clinical and business use cases. Implementing platforms that support deployment of multiple AI applications for medical imaging across the clinical environment, these providers are demonstrating that AI in radiology can unlock value downstream, such as in oncology, neurology or surgery.

AI will have a significant impact on the quadruple aim. First, AI will ensure that busy clinicians are better equipped to deal with increasingly complex imaging studies and can effectively communicate results to referring physicians. Second, by fostering earlier and more appropriate interventions, AI can reduce costs for payers and ensure that patients have a greater understanding of their conditions and achieve better outcomes. Third, by detecting conditions earlier, during routine scans, and allowing for better characterization of conditions before a patient ever becomes ill, and finally, AI could address the challenges of improving the health and well-being of populations around the world.

Going forward, it is important to look at the big picture and understand how efficient, effective medical imaging that leverages AI can improve many outcomes along the healthcare continuum. 

Original post here