Assessment of AI/ML in digital pathology | US Food and Drug Administration (FDA)

As the cornerstone of modern medicine and cancer care, pathology is undergoing a revolutionary change from reviewing glass slides of tissue specimen under a microscope to displaying scanned digital whole-slide images (WSI) on monitors. The availability of large amount of WSI data and the rise of artificial intelligence (AI) and machine learning (ML) technologies have given rise to the field of AI in digital pathology, where AI/ML algorithms are developed for cancer detection, segmentation and quantification of cells or other structures of interest, and so on. The overall goal of this project is to develop assessment methodologies that enables collecting evidence of safety and effectiveness in a least burdensome fashion thereby facilitating the deployment of AI/ML technologies to pathology applications to improve diagnostic accuracy, efficiency, and patient management. We leverage the use of publicly available datasets and software code and develop image processing and AI/ML pipelines. We investigate study design and data analysis methods for characterizing the generalizability of AI/ML performance and its robustness to image acquisition variations due to e.g., scanner, lab, and site. The REU student will gain experiences in some of the following areas: data collection and management, programming, image processing and statistical analysis methodologies.

Duties: The student will gain experience in image segmentation, analysis, management, and artificial intelligence algorithms in imaging applications