The field of precision oncology relies on the examination of tissue pathology, increasingly augmented by machine learning techniques facilitated by digital technologies. Targeted cancer therapies face challenges posed by the intratumoral heterogeneity of biomarker expression, while anti-cancer immunotherapies demand an evaluation of immune cells within the spatial context of the tissue microenvironment. Various computational methods have been proposed to analyze these mostly imperceptible features, aiming to enhance prognostic and predictive disease models.

This presentation will focus on the potential of hexagonal tiling analytics, utilizing data generated through digital image analysis to scrutinize spatial aspects of the tumor microenvironment. Firstly, a hexagonal grid-based systematic subsampling of digital immunohistochemistry data allows for the quantification of intratumoral heterogeneity of biomarker expression. This is achieved by computing Haralick’s entropy and other texture indicators, often surpassing the performance of biomarker expression rates alone in prognostic models. Secondly, hexagonal tiling facilitates the extraction of the tumor-host interface zone, preserving its spatial directionality to compute density distributions of immune cells across the zone, conceptualized as an immunogradient. Both computational models have been reported to yield independent prognostic indicators in various solid tumors.
 

Several new AI-based digital cytology systems have emerged to provide computer-assisted screening of Pap tests. This talk will review the historical background, innovative technology of digital cytology and compare several of the newer systems on the market. The speaker will share his experience validating one of these systems for use in clinical practice. Also, the talk will highlight the benefits of deploying digital Pap tests in practice and present the recent recommendations published by the American Society of Cytopathology for AI use in cytology.

Aiforia provides powerful deep-learning AI and cloud-based technology to pathologists and scientists across clinical, pre-clinical, and academic labs. This presentation will showcase our recent clinical advancements, particularly in breast and prostate cancer. Additionally, we'll introduce Aiforia Create, our user-friendly research platform for AI training. Its advanced features allow pathologists to easily create complex AI models to enhance their work.