Multi-source Domain Adaptation Using Gradient Reversal Layer for Mitotic Cell Detection

TL;DR

This paper presents a method for mitotic cell detection that employs multi-source domain adaptation with a gradient reversal layer, aiming to improve generalization across different datasets in the MIDOG 2021 challenge.

Contribution

The paper introduces a novel multi-source domain adaptation approach using a gradient reversal layer specifically for mitotic cell detection.

Findings

Improved detection accuracy across multiple datasets.

Effective domain generalization demonstrated in MIDOG 2021 challenge.

Method outperforms baseline models in mitotic cell detection.

Abstract

This is a write-up of our method submitted to Mitosis Domain Generalization (MIDOG 2021) Challenge held in MICCAI2021 conference.