Digital pathology software company Proscia said it has released the first in a series of artificial intelligence applications designed to advance the practice of pathology.
A module on Proscia’s Concentriq platform, DermAI uses deep learning to read and automatically classify hundreds of variants of skin diseases into pre-diagnostic categories to help reduce costly errors and improve laboratory quality and efficiency.
Unlike other parts of the healthcare industry, pathology labs have remained dependent on manual, time-consuming processes to track key business metrics, Proscia chief product officer Nathan Buchbinder recently told Medical Design & Outsourcing in an interview. This has kept labs from improving productivity and making informed decisions that could otherwise improve the bottom line, especially in light of decreasing reimbursements, Buchbinder said.
Philadelphia-based Proscia said it trained and tested the DermAI algorithm using patient biopsies from academic and commercial dermatology laboratories, including Cockerell Dermatopathology, Dermatopathology Laboratory of Central States, University of Florida, and Thomas Jefferson University Hospital. This multi-site study successfully validated the performance of DermAI using more than 20,000 patient biopsy slides, according to the company. Proscia intends to submit DermAI to the FDA for review to use in clinical diagnosis.
“To date, attempts to apply AI to pathology have been engineered in isolated development environments using toy datasets,” said Proscia CEO David West in a news release. “The challenge in fulfilling the promise of deep learning in diagnostic medicine is bringing to market a solution that can perform in the real world where we face tremendous variability among labs, systems and specimens.”