Contrastive Learning for Mitochondria Segmentation

TL;DR

This paper introduces a contrastive learning framework that improves mitochondria segmentation in electron microscopy images by focusing on hard examples and pixel-wise contrastive loss, leading to better feature representation and segmentation accuracy.

Contribution

The paper proposes a novel contrastive learning approach with pixel-wise loss and hard example sampling for enhanced mitochondria segmentation.

Findings

Achieves better or comparable results to state-of-the-art methods on MitoEM and FIB-SEM datasets.

Demonstrates that pixel-wise contrastive loss improves feature discrimination.

Shows that hard example sampling enhances segmentation performance.

Abstract

Mitochondria segmentation in electron microscopy images is essential in neuroscience. However, due to the image degradation during the imaging process, the large variety of mitochondrial structures, as well as the presence of noise, artifacts and other sub-cellular structures, mitochondria segmentation is very challenging. In this paper, we propose a novel and effective contrastive learning framework to learn a better feature representation from hard examples to improve segmentation. Specifically, we adopt a point sampling strategy to pick out representative pixels from hard examples in the training phase. Based on these sampled pixels, we introduce a pixel-wise label-based contrastive loss which consists of a similarity loss term and a consistency loss term. The similarity term can increase the similarity of pixels from the same class and the separability of pixels from different classes in feature space, while the consistency term is able to enhance the sensitivity of the 3D model to changes in image content from frame to frame. We demonstrate the effectiveness of our method on MitoEM dataset as well as FIB-SEM dataset and show better or on par with state-of-the-art results.