A Novel AI based prognostic approach using histopathology images for Hodgkin Lymphoma

Abstract

Hodgkin Lymphoma is a type of tumor originating from a particular white blood cell type called lymphocyte. It is marked by the presence of multinucleated Reed-Sternberg cells (RS cells) which are found in the lymph nodes of the patient. It accounted for 0.5% of all the newly reported cancer cases and 0.1% of all cancer-related deaths in 2023, which makes it a rare cancer type, but still is very relevant for study. This study applied various approaches on histopathological images to identify RS cells along with the microenvironment cells in order to predict whether the patient is suffering from Hodgkin’s Lymphoma and also predict the type of microenvironment for the patient. The data was obtained from Tata Memorial Hospital, Mumbai. The study initially predicted the results on the Hodgkins dataset after getting trained on the online histopathological open-source dataset. Due to issues related to training on an open-source dataset, it was finally trained on a manually labeled dataset annotated by an expert annotator who individually annotated the patches from the whole slide images to obtain the required cells. Therefore, the study describes an attempt to create a pipeline in order to automate the process of segmentation and classification of cells in a given Whole Slide Image to predict if the patient is at an advanced stage of Hodgkin’s Lymphoma and also predict the tissue microenvironment.