Almost every patient story starts with an image

How AI-augmented diagnostic imaging is driving new approaches to collaborative, patient-specific precision care

We all know the old saying that a “picture is worth a thousand words.” In medicine, this saying comes to life every day. As noted in a 2012 study published in the Journal of the American Medical Association, the authors found that more than 80 percent of all hospital and health system visits included at least one diagnostic imaging study related to more than 23,000 different conditions. And we know that in many of these clinical pathways, diagnostic imaging is not only used, but is used very early in the determination of what to do clinically for the patient.

When you combine the importance of diagnostic imaging, the ever-increasing volume of patients and, thus, radiology exams, staffing shortages, and burnout among radiologists and other clinicians, the need for technology to enhance the process of diagnostic imaging has never been greater. This is where Artificial Intelligence (AI) for diagnostic imaging comes into play. The use of AI in imaging started way back in the early 1990s when AI-powered speech recognition came to radiology. Today, over ninety percent of all radiology reports are produced using advanced AI technology. Since 2017, the development and adoption of pixel-based AI for diagnostic imaging has rapidly progressed as the technology has improved and meaningful use cases have been developed.

Currently there are approximately 200 FDA cleared pixel-based AI models helping radiologists and other clinicians in multiple areas. For example, workflow triage models quickly identify and prioritize time-critical cases such as potential intra-cranial hemorrhage. Other models automate time-consuming but important tasks such as assessing breast density, measuring and characterizing potentially cancerous lung nodules, or auto populating fields in the radiology report with data that radiologists otherwise would need to enter manually. These AI technologies are enabled by the increased adoption of Common Data Element (CDE) standards in radiology. The implementation of CDE standards allows health systems or anyone else that has a reporting system to unlock value from vast amounts of diagnostic imaging and radiology reporting data.  For example, structured data can automatically prompt the inclusion of evidence-based clinical guidance and trigger communication between radiologists and referring physicians to reduce the incidence of failed follow-up and adverse outcomes.

The sheer volume and importance of diagnostic imaging in nearly every area of medicine also prompted radiologists and other health care stakeholders to explore ways to partner to extend the benefits of AI beyond radiology reading rooms at scale to deliver better patient outcomes across the care continuum. What has recently emerged is an AI-powered cloud platform that can facilitate the sharing of diagnostic images across disparate sites and deliver patient-specific data and insights from these images into existing clinical and administrative workflows across the healthcare ecosystem to promote better patient care, lower costs and enhance physician efficiency. From patient screening to diagnosis, treatment and follow up, this open cloud platform can facilitate the extraction and delivery of these AI-enhanced insights among radiologists, other clinicians, life science companies, health insurers and self-insured employers to accomplish the following objectives:

• For Healthcare Provider Organizations: apply AI-generated insights from diagnostic imaging to support earlier diagnosis, inform treatment options and planning, and augment clinical decision-making. There are many examples of these use cases including in pulmonology. For example, by offering AI-generated insights, one local hospital was easily able to screen and quantify patients for their Bronchial Valve Program. With AI, patients could be easily identified through LCS programs and diseases like COPD could be detected much earlier. This reduces the cost of care for patients and supports Value Based Care initiatives as well.
• Health Plans & Self-Insured Employers: apply AI-generated insights from diagnostic imaging to improve payer-provider collaboration to help address quality, enhance care coordination and improve the prior authorization process.
• Life Science Companies: apply AI-generated insights from diagnostic imaging to better inform the use of precision therapeutics and companion diagnostics by qualifying patients eligible for life-enhancing treatment options and clinical trials more quickly and efficiently.

For example, Nuance, Olympus, and AI developer Imbio are working with radiologists and pulmonologists to determine if patients with severe emphysema, a form of chronic obstructive pulmonary disease, are eligible for precision treatment with surgically implanted lung valves that can significantly improve quality of life. Radiologists use an AI model to obtain key measurements needed to qualify new or existing patients in collaboration with their pulmonologists. Olympus can then use the data to provide devices tailored for each patient for surgical implantation.

The goal of AI-augmented precision imaging is to improve patient outcomes while addressing systemic inefficiencies that contribute to physician burnout, poor health care experiences, and higher costs. Although pixel-based AI is still relatively young, deployments are increasing, meaningful outcomes are being demonstrated, and stakeholder interest is rapidly accelerating as the health care industry looks to drive better patient care, improve efficiencies, and overcome staff shortages to ensure financial integrity during the pandemic and in an eventual post-Covid world.

For more information, visit www.nuance.com.

Peter Durlach is the Executive Vice President and Chief Strategy Officer for Nuance Communications where he holds a pivotal role in advancing the portfolio of AI-powered solutions to align with industry pressures and shifting needs of customers.