Self-supervised dense representation learning for live-cell microscopy with time arrow prediction

TL;DR

This paper introduces a self-supervised learning approach for live-cell microscopy that predicts the temporal order of image regions to learn detailed representations, improving cell detection and segmentation with limited annotations.

Contribution

The novel method uses time arrow prediction for self-supervised dense representation learning in microscopy, outperforming supervised methods especially with scarce labeled data.

Findings

Outperforms supervised methods in cell detection and segmentation

Captures time-asymmetric biological processes at pixel level

Effective with limited annotated training data

Abstract

State-of-the-art object detection and segmentation methods for microscopy images rely on supervised machine learning, which requires laborious manual annotation of training data. Here we present a self-supervised method based on time arrow prediction pre-training that learns dense image representations from raw, unlabeled live-cell microscopy videos. Our method builds upon the task of predicting the correct order of time-flipped image regions via a single-image feature extractor followed by a time arrow prediction head that operates on the fused features. We show that the resulting dense representations capture inherently time-asymmetric biological processes such as cell divisions on a pixel-level. We furthermore demonstrate the utility of these representations on several live-cell microscopy datasets for detection and segmentation of dividing cells, as well as for cell state classification. Our method outperforms supervised methods, particularly when only limited ground truth annotations are available as is commonly the case in practice. We provide code at https://github.com/weigertlab/tarrow.