Johns Hopkins School of Medicine, Proscia team to advance AI in pathology

Register now

Johns Hopkins School of Medicine is working with Proscia, a vendor of advanced digital pathology software, to develop computational applications for the specialty in addressing several types of diseases and specialties.

The initiative aims to incorporate artificial intelligence to advance the practice of pathology for multiple diseases, with the ultimate aim of making accurate AI applications that laboratories, physicians and patients can rely on.

Separately, the vendor also is working with several other healthcare and educational organizations in partnerships; these include the University of Florida, Cockerell Dermatopatholgy, the Dermatopatholgy Laboratory of Central States and Thomas Jefferson University Hospital, on various initiatives regarding digital pathology.

Johns Hopkins has a large supply of digital images and their related diagnoses, and it is sharing with Proscia its library of diagnosed cases and related digital pathology images.

To make an AI-based system effective, it must be trained. In the case of pathology, the AI training materials are digital images of tissue samples, which are digital images that are made when a physician takes a biopsy that’s prepared on glass slides that are scanned to create digital images—related information includes the diagnostic report related to each set of digitized slides.

Training a successful AI system for pathology requires diverse, high-quality pathology data. Diverse data helps ensure an AI system is accurate across a wide variety of diseases, methods of biopsy, preparation of tissue, tissue staining procedures and digital scanning processes. Further, the greater variation of data used to train the AI application, the more accurate the applications will be when analyzing cases coming from other physicians and labs, notes Michael Bonham, MD, chief medical officer at Proscia.

“As digital technology continues to gain traction, AI-driven applications will advance this adoption by driving economic and clinical benefits,” Bonham adds. “Disease-specific AI applications can bring efficiency, productivity and quality to support tissue diagnosis, overcome subjectivity inherent in pathology and address the shortage of pathologists.“

Also See: AI-based software tool could help pathologists identify cancer cells

The impetus behind the collaboration was the desire to advance the field of pathology and apply artificial intelligence technology to it, says David West, CEO at Proscia. “Johns Hopkins came to the field to apply AI to pathology, which at this time is still built on brick and mortar using glass slides and microscopes,” he explains.

Pathology labs are starting to shift from glass to digital images, and some early adopters are going digital and adopting digital pathology scanners that are beginning to mature in their technology.

For now, however, most pathologists are not using algorithms as they continue to use visual interpretations to make their diagnoses and treatment decisions. In part, that’s because current algorithms need to mature and pathologists need time to deploy mature algorithms into their workflows.

For reprint and licensing requests for this article, click here.