Digital pathology: Transforming diagnosis and improving patient care

“Digital pathology is poised to dramatically change the way tissue samples are analyzed for diagnosis, increasing collaboration among colleagues, and harnessing a vast array of expertise to improve patient care,” according to Darren Wheeler, MD, vice president, Pathology and Medical Services at Quest Diagnostics. In a recent episode of the podcast, Diagnostics Dialogues, Dr Wheeler provided an overview of the enormous potential that digital tools can pose in transforming the practice of clinical pathology, accelerating, and enhancing the overall diagnostic process.

Biopsies are crucial for diagnosis of many types of cancer, and other diseases as well. A tissue sample is typically stained with dyes or antibodies for microscopic analysis, and may be genetically analyzed as well. It’s the pathologist’s job to put all that information together to render a diagnosis. But no pathologist is an expert on every type of cancer, Dr Wheeler noted, and consulting with colleagues is the norm in the field. In the past, that has required sending the slides themselves, which might be forwarded to yet another expert if the case is unusual. “This could take a week or more to get back the opinion you are seeking,” Dr Wheeler said.

That all changes when the images from the slide are digitized. “Within minutes, you can share that digital image across a network of pathologists,” no matter where they are located, who can collaboratively analyze and discuss the same image in real-time, rendering a diagnosis quickly and with a high degree of certainty. “That can improve patient care,” Dr Wheeler said.

Pathologists improve their diagnostic acumen by examining thousands of slides over their careers. With digital imagery, an artificial intelligence (AI) program can examine thousands of images in the course of a day, and leverage the expertise of pathologists to learn to distinguish different tumors and disease stages. AI is just beginning to enter the clinical pathology world, but Dr Wheeler said it is likely to dramatically improve the pathologist’s capabilities.

“There is always human error in healthcare,” Dr. Wheeler said, and at the end of a long day, the pathologist may miss a subtle or ambiguous sign of pathologic change. “But the computer never gets tired. This is extremely helpful, to make sure some of these smaller things aren't missed.”

“We are not letting the computer make the diagnosis on its own,” he added, but instead the AI program works with the pathologist, screening images and making provisional conclusions that the pathologist then can accept or reject. “Ultimately it is still the pathologist who is making the diagnosis.”

But when the program offers a conflicting opinion, “that might be helpful to me,” Dr Wheeler said, given how powerful the algorithms are and the breadth of cases they learn from. “That might make me think this is a case that I should take to another expert and get their opinion. It’s like having someone looking over your shoulder and steering you in the right direction.”

Except that, given the large universe of diagnosed cases the program will be trained on, it’s more like “the expertise of 10,000 experts in that same disease state is going to be looking over my shoulder,” Dr Wheeler said. “And I think that is going to not only improve the accuracy and quality of diagnosis, but it's going to bring forward insights that we’ve never had before,” such as treatment recommendations based on the unique characteristics of tumor and patient, “creating a comprehensive cancer report that includes all this information that can inform therapy.” And the same strategy can be brought to bear on other diseases, such as Alzheimer’s disease or osteoporosis. And this comprehensive approach will ultimately lead to better, and more equitable, treatment for all patients.