Applying AI in reading pathology slides promises to significantly improve the efficiency and accuracy of pathology labs.

Roche and Bristol Myers Squibb (BMS) announced today a collaboration that seeks to aid the development of two clinical trial assays via the development and deployment of digital pathology algorithms. The data from both projects will be used to improve cancer diagnostics to pinpoint personalized treatments for cancer patients, the two companies said.

In the first project, Roche Digital Pathology will create an image analysis algorithm leveraging artificial intelligence to aid pathologist in interpreting the existing VENTANA PD-L1 (SP142) Assay. BMS will, in turn, use the newly developed algorithm to generate biomarker data from clinical trial samples.

The second project will leverage Roche’s Open Environment collaboration with digital pathology firm PathAI announced in October. This collaboration is enabled by Roche’s Digital Pathology Open Environment, which allows pathologists to securely access third-party AI-powered technology alongside Roche’s growing menu of AI-based image analysis tools. In this case a PathAI-developed algorithm for the CD8 biomarker will be used by BMS to analyze clinical trial samples that have been stained with Roche’s CD8 assay to generate quantitative spatial biomarker data.

“The Bristol Myers Squibb and PathAI collaborations are among the first examples where AI technology and digital pathology applications are playing a role in developing treatments for patients,” said Jill German, head of Roche Diagnostics Pathology Customer Area in a press release announcing the collaboration. “By using our NAVIFY Digital platform to interpret tissue based assays and AI algorithms, pathologists are better able to identify targeted therapy options, ultimately improving patient care.”

The field of digital pathology, whereby whole slide images are captured and stored electronically, is quickly being transformed by AI applications. The same technology that allows your face to unlock access to your cell phone is being applied to recognizing and quantifying specific cell types—and their locations—from digital pathology images. The promise of AI-based algorithms providing analysis of tissue samples for the detection of cancer and the identification of new biomarkers promises to have a significant impact on the practice of precision oncology. It can also significantly improve workflow within a pathology lab by automating the process of tissue analysis.

“We believe digital methods will bring significant improvements in standardization and interpretation of tissue-based assays and will enable broader access to tissue based assays. The ability to more deeply interrogate images will present opportunities to better understand disease biology, potentially leading to expanded and improved drug development options and ultimately highly effective patient selection strategies,” said Sarah Hersey, vice president, Translational Sciences and Diagnostics at BMS.

For PathAI, the work with Roche on the two projects is merely the latest in a growing roster of researcher partnerships and collaborations the company has announced in the past year and is another proving ground for it’s AI-powered image analysis algorithms.

In November, at the Society for Immunotherapy of Cancer meeting, the company released new results from an exploratory biomarker analysis on digitized slides that applied AI-predicted CD8 topology assessments of patients with advanced melanoma. Presenting data in two posters at the meeting and using data form the CheckMate 067 clinical trial, PathAI was able to demonstrate that PD-L1 negative CheckMate 067 patients who were also CD8+ in specific subregions of tumors had overall survival benefit when treated with ipilimumab plus nivolumab combinations compared to those that were treated with ipilimumab monotherapy.

The posters presented revealed the promised of machine learning (ML) approaches as a prognostic tool for the selection of appropriate targeted therapies for patients and highlighted “the promise of ML-based CD8+ phenotype scoring to reveal populations of patients that might have the greatest benefit from existing treatments,” the company noted.

This site uses Akismet to reduce spam. Learn how your comment data is processed.