Computational pathology for breast cancer: Where do we stand for prognostic applications?

Abstract

The very early days of artificial intelligence (AI) in healthcare are behind us. AI is now spreading in the healthcare sector and is gradually being implemented in routine clinical practice. Driven by the increasing digitization of microscope slides, computational pathology (CPath) is further strengthening the role of AI in the field of oncology. CPath is transforming fundamental research as well as routine clinical practice, both for diagnostic and prognostic applications. In breast cancer, CPath holds the potential to address several unmet clinical needs, particularly in the areas of biomarkers and prognostic tools. Indeed, multiple applications are on their way, ranging from predicting clinically meaningful endpoints to offering alternatives to gene-expression testing and detecting molecular alterations directly from digitized whole slide images. However, to fully harness the potential of CPath, several challenges must be overcome. These include improving the availability of multimodal patient data, advancing the digitalization of pathology laboratories, increasing adoption within the medical community, and navigating regulatory hurdles. This review offers an overview of the current landscape of CPath in breast cancer, highlighting the progress made and the hurdles that remain for its widespread clinical adoption in prognostic applications.