July 24, 2023 — PathAI, a leading provider of AI-powered pathology tools to advance precision medicine, today announced the availability of AIM-HER2 Breast Cancer [i] for research use by clinical laboratories, researchers, and drug developers. AIM-HER2 Breast Cancer delivers automated digital HER2 scoring and is the market’s first algorithm to use additive multiple instance learning (aMIL) heatmaps to visualize the slide features driving the algorithm’s predicted score. AIM-HER2 will be available to biopharma researchers and clinical research lab pathologists on PathAI’s AISight [i] digital pathology platform in both the clinical research lab and clinical trial setting.
“HER2 IHC assays have provided immense impact to patients by allowing for a widely available test to establish drug eligibility,” said Mike Montalto, Chief Scientific Officer at PathAI. “With AIM-HER2 Breast Cancer, we at PathAI sought to build upon the success of HER2 testing by assisting pathologists in their ability to more confidently score HER2, especially in borderline cases that can be the most challenging and time consuming to review. We also look forward to partnering with drug developers who may be interested in enhancing the scoring of the more recently described HER2 low assay as an important emerging patient population.”
In addition to predicting slide-level HER2 score, AIM-HER2 Breast Cancer’s visualizations on AISight utilize additive multiple instance learning (aMIL) heatmaps to deliver a more interpretable and explainable prediction, diminishing the ‘black box’ challenges typically associated with understanding how AI predictions are made.
“Our interpretable heatmaps are critical to driving utilization and adoption of AI – because it’s not about blind trust,” said Eric Walk, Chief Medical Officer at PathAI. “The results can be interpreted, explained – and confirmed – by humans. It also streamlines workflow as it allows pathologists to hone in on specific features to confirm the algorithm score and output vs. needing to analyze the entire slide.”
AIM-HER2 Breast Cancer was developed using 157,000 tissue annotations and consensus scores on a dataset of over 4,000 slides (collected from more than 65 expert breast pathologists). In addition to slide-level HER2 scoring, AIM-HER2 Breast Cancer quantifies invasive carcinoma and provides a comprehensive analysis of the entire WSI without necessitating manual selection of the region of interest (ROI).
For more information: www.PathAI.com
Footnote
[i] Both AISight and the AIM-HER2
Breast Cancer Algorithm
are intended for research-use only. Not for use in diagnostic procedures.
May 03, 2024 
