Philips, PathAI team up on breast cancer diagnostic

Philips headquarters
Philips and PathAI aim to create an artificial intelligence-based solution that could relieve the burden of increasing cancer caseloads falling on a shortage of pathologists.

Philips is partnering with PathAI, which is developing deep-learning algorithms for the detection and diagnosis of disease, to improve the speed and accuracy of cancer diagnosis. Their first focus will be breast cancer.

The pair aims to apply deep learning to “massive” pathology datasets to inform diagnosis and treatment decisions, Philips said in a statement Wednesday. They will first develop deep-learning applications to automatically spot cancerous lesions in breast cancer tissue.

This application could ease the burden on pathologists, who have historically reviewed tumor tissue samples manually. While tumor analysis is essential for diagnosis, doing it manually is time-consuming and may put pressure on pathologists to read and analyze slides more quickly, Philips said.

"Breast cancer is the most common cancer in women worldwide, with over 250,000 new cases diagnosed every year in the U.S.," said PathAI CEO Andy Beck, in the statement.

"[Identifying] the presence or absence of cancer in lymph nodes is a routine and critically important task for a pathologist. However, it can be extremely laborious using conventional methods. Research indicates that pathologists supported with computational tools could be both more accurate and faster," he said

Philips’ Illumeo platform uses adaptive intelligence to help radiologists work more efficiently, while its IntelliSite Pathology Solution is an automated digital pathology system that includes a slide scanner, image management system and software tools. Last June, the company bought Northern Ireland-based PathXL, which focuses on image analysis and digital pathology.

A number of players are applying artificial intelligence to diagnose breast cancer from medical imaging, including Samsung, which used a deep learning algorithm to detect breast cancer lesions in ultrasound images. It has been challenging to create computer-aided diagnostic solutions for ultrasound because it is read in real time and ultrasound images have a lower resolution and more noise than other types of imaging, such as MRI.

Meanwhile, researchers from Houston Methodist have developed software that predicts breast cancer risk from patient charts and mammograms, and a Harvard-MIT team have applied AI to the diagnosis of breast cancer from slides of lymph node cells.