Learning under Distributed Weak Supervision

TL;DR

This paper explores using crowdsourced weak annotations from non-experts to train neural networks for fetal brain segmentation in MRI, offering a scalable alternative to traditional expert labeling.

Contribution

It introduces a method leveraging crowdsourcing for weak supervision in medical image segmentation, reducing reliance on expert annotations.

Findings

Crowdsourced annotations enable effective training of segmentation models.

The approach achieves promising results compared to fully supervised methods.

It addresses scalability issues in medical image annotation.

Abstract

The availability of training data for supervision is a frequently encountered bottleneck of medical image analysis methods. While typically established by a clinical expert rater, the increase in acquired imaging data renders traditional pixel-wise segmentations less feasible. In this paper, we examine the use of a crowdsourcing platform for the distribution of super-pixel weak annotation tasks and collect such annotations from a crowd of non-expert raters. The crowd annotations are subsequently used for training a fully convolutional neural network to address the problem of fetal brain segmentation in T2-weighted MR images. Using this approach we report encouraging results compared to highly targeted, fully supervised methods and potentially address a frequent problem impeding image analysis research.