Abstract: Histopathology slides are routinely evaluated for the diagnosis of breast cancer and likely contain significant untapped biological signal. We created a pipeline that combines omics and image analyses and discovered a novel concordance between a patient’s hormone receptor status and immune cell infiltration in their tumors. Furthermore, the image-based classifier was shown to identify the features associated with lymphocytes. Our method can be extended to investigate other diseases.
Learning Objective 1: After reading the poster, the reader should be better able to:
* Understand the motivation and methods behind the integrative omics/image analysis.
* Appreciate the opportunities and challenges in automated histopathology analysis.
* Become familiar with machine-learning applications of breast cancer pathology, histomorphology and subtyping.
William Yuan (Presenter)
Harvard Medical School
Isaac Kohane, Harvard Medical School
Kun-Hsing Yu, Harvard Medical School