Machine Learning

The goal of this session is to educate attendees on methods of pathological assessment of human cancers in current clinical practice and introduce novel approaches to human tissue analysis that may impact cancer pathology in the future. Presentations by clinically trained pathologist investigators will encompass both current practice in diagnostic pathology of neoplasia as well as emerging technologies in human tissue analysis. Each presentation will include concepts related to the diagnosis, classification, and/or staging of human tumors, as well as discussion of novel approaches that are likely to be integrated into clinical practice, including digital image analysis, multi-plex tissue profiling, and three-dimensional mapping.

Ki67 labeling index (LI) is the preferred method of establishing grade for gastroenteropancreatic neuroendocrine tumors (GEP-NETs). There are a number of acceptable methods for calculating the Ki67 LI in GEP-NETs, but application of quantitative image analysis is a promising method to increase accuracy and repeatability. KiNet is a single-stage deep-learning-based detection and classification pipeline for Ki67 LI with performance meeting or exceeding the current state of the art.

Cell or nucleus detection is a fundamental task in microscopy image analysis and has recently achieved state-of-the-art performance by …

Advances in artificial intelligence (AI), especially in deep learning, improve pathological image analysis in basic, translational, and …

PURPOSE: We focus on the problem of scarcity of annotated training data for nucleus recognition in Ki-67 immunohistochemistry …

OBJECTIVE: Nucleus recognition is a critical yet challenging step in histopathology image analysis, for example, in Ki67 …