DASGAN -- Joint Domain Adaptation and Segmentation for the Analysis of Epithelial Regions in Histopathology PD-L1 Images

TL;DR

This paper presents DASGAN, a deep learning model that jointly performs domain adaptation and segmentation of epithelial regions in histopathology images, enabling accurate tumor epithelium analysis across different staining domains without serial sectioning.

Contribution

The novel DASGAN framework combines domain adaptation and segmentation using unpaired image translation, reducing annotation effort and enabling PD-L1 positive/negative region differentiation.

Findings

Achieves high segmentation accuracy compared to state-of-the-art methods.

Effectively differentiates PD-L1 positive and negative tumor epithelium.

Enables automated PD-L1 Tumor Cell scoring.

Abstract

The analysis of the tumor environment on digital histopathology slides is becoming key for the understanding of the immune response against cancer, supporting the development of novel immuno-therapies. We introduce here a novel deep learning solution to the related problem of tumor epithelium segmentation. While most existing deep learning segmentation approaches are trained on time-consuming and costly manual annotation on single stain domain (PD-L1), we leverage here semi-automatically labeled images from a second stain domain (Cytokeratin-CK). We introduce an end-to-end trainable network that jointly segment tumor epithelium on PD-L1 while leveraging unpaired image-to-image translation between CK and PD-L1, therefore completely bypassing the need for serial sections or re-staining of slides. Extending the method to differentiate between PD-L1 positive and negative tumor epithelium regions enables the automated estimation of the PD-L1 Tumor Cell (TC) score. Quantitative experimental results demonstrate the accuracy of our approach against state-of-the-art segmentation methods.