This website requires JavaScript.

Weakly-Supervised Deep Learning Model for Prostate Cancer Diagnosis and Gleason Grading of Histopathology Images

Mohammad Mahdi BehzadiMohammad MadaniHanzhang Wang ...+5 Sheida Nabavi
Dec 2022
Prostate cancer is the most common cancer in men worldwide and the secondleading cause of cancer death in the United States. One of the prognosticfeatures in prostate cancer is the Gleason grading of histopathology images.The Gleason grade is assigned based on tumor architecture on Hematoxylin andEosin (H&E) stained whole slide images (WSI) by the pathologists. This processis time-consuming and has known interobserver variability. In the past fewyears, deep learning algorithms have been used to analyze histopathologyimages, delivering promising results for grading prostate cancer. However, mostof the algorithms rely on the fully annotated datasets which are expensive togenerate. In this work, we proposed a novel weakly-supervised algorithm toclassify prostate cancer grades. The proposed algorithm consists of threesteps: (1) extracting discriminative areas in a histopathology image byemploying the Multiple Instance Learning (MIL) algorithm based on Transformers,(2) representing the image by constructing a graph using the discriminativepatches, and (3) classifying the image into its Gleason grades by developing aGraph Convolutional Neural Network (GCN) based on the gated attentionmechanism. We evaluated our algorithm using publicly available datasets,including TCGAPRAD, PANDA, and Gleason 2019 challenge datasets. We also crossvalidated the algorithm on an independent dataset. Results show that theproposed model achieved state-of-the-art performance in the Gleason gradingtask in terms of accuracy, F1 score, and cohen-kappa. The code is available at