Hologic took a giant step in detecting subtle potential cancers in breast tomosynthesis images. The Marlborough, MA-based company received clearance for the company’s Genius AI Detection technology, a new deep learning-based software.
The company said the new software delivers key metrics at the time of image acquisition to help radiologists categorize and prioritize cases by complexity and expected read time in order to optimize workflow and expedite patient care.
“Consider this one-day workflow: After the acquisition of patient images, Genius AI Detection provides results to the technologist in less than a minute,” Jennifer Meade, Division President, Breast & Skeletal Health Solutions, Hologic, told MD+DI. “Based on the case score, the algorithm flags a percentage of cases as having a higher level of concern. If a patient’s exam is flagged, the provider may choose to prioritize the case for immediate reading and to retain the patient while the exam is being reviewed. If, after radiologist review, additional imaging is required, there is the possibility to perform that as part of the same visit, as opposed to scheduling a call-back appointment.”
Genius AI Detection sensitivity was shown to be 93% to 94% and specificity was shown to be 37% to 41% in standard resolution and high-resolution modes.
The company said it will eventually file to get the software approved in Europe.
“Hologic actually has a long history of machine learning expertise that goes back to 1998 when our first machine learning-based CAD product to detect cancers from mammographic images, ImageChecker CAD, was approved by FDA,” Meade said. “Since then, Hologic has been an industry leader in developing CAD support tools that use machine learning techniques.”
She added, “Building upon these advancements in machine learning, Hologic has been working to build capabilities to employ deep learning. To develop Genius AI Detection, Hologic leveraged its large installed base of tomosynthesis systems (Hologic’s 3Dimensions and Selenia Dimensions systems) to collect the necessary training data for the development of the models to detect cancers in tomosynthesis images."
Hologic has had a resurgence of late. The company brought in more than $1.3 billion in its most recent reported earnings quarter – a direct result of strong sales from the firm’s COVID-19 offerings. Specifically, Hologic said its molecular diagnostics division, which includes its COVID-19 tests, brought in $819 million in revenue for the quarter a 376% increase compared to the final quarter of 2019.