A Dilated Residual Hierarchically Fashioned Segmentation Framework for Extracting Gleason Tissues and Grading Prostate Cancer from Whole Slide Images.

Prostate cancer (PCa) is the second deadliest form of cancer in males. PCa severity can be clinically graded by examining the structural representations of Gleason tissues. The paper proposes a framework for segmenting Gleason tissues and grading PCa using Whole Slide Images (WSI). Our approach encompasses two main contributions: 1) An asymmetric dilated residual segmentation model integrating a novel hierarchical decomposition scheme to extract textured Gleason tissues. 2) A three-tiered loss function to ensure accurate recognition of the cluttered regions in the cancerous tissues. The proposed framework has been extensively evaluated on a large-scale PCa dataset containing 10,516 whole slide scans (with around 71.7M patches), where it outperforms state-of-the-art schemes in several metrics for extracting the Gleason tissues and grading the progression of PCa.