Weakly supervised learning segments kidney structures in microCT scans
Weakly supervised neural network: segmentation of complex structures in X-ray microCT
Artificial IntelligenceComputer Vision and Pattern Recognition
Summary
It is hard and expensive to fully label medical images to find tiny parts inside organs. The authors tested a way to teach a computer to find these parts using only a few dots to show where they are, instead of full drawings. They taught a neural network this way to find small kidney structures in X-ray images of rat kidneys. Their method worked well, and using a few full drawings made it even better, almost as good as fully labeled training. This could help scientists analyze complex X-ray images faster and with less manual work.
What this means in practice
- •For medical imaging developers: Improve kidney structure segmentation in tomographic scans with fewer manual annotations by using sparse dot labeling combined with selective full annotations.
- •For industrial nondestructive testing teams: Adapt weak supervision methods to identify small, complex features in X-ray scans of manufactured parts, reducing the need for detailed labeling.
Tested on one dataset.
Authors
Daniele Rusconi, Michela Ascolese, Stephanie Fest-Santini, Alberto Bravin, Maurizio Santini
Abstract
Segmentation of complex structures in X-ray tomographic data is a fundamental task in biomedical research, but it often requires large amounts of precisely annotated data, making fully supervised approaches costly and difficult to scale. In this study, weakly supervised deep learning is investigated as a strategy to reduce annotation effort while maintaining accurate segmentation. A two-dimensional convolutional neural network based on the nnU-Net framework was adapted to a weak supervision setting using sparse dot-based annotations, complemented by a limited number of fully segmented images. The approach was evaluated on high-resolution microCT slices of rat kidneys, targeting the segmentation of renal glomeruli, which are small, low-contrast anatomical structures. Results indicate that weak supervision provides a meaningful learning signal, enabling reliable localization of glomeruli even in the absence of dense labels. Incorporating a small set of high-quality annotations substantially improves segmentation performance, approaching that of a fully supervised model. These findings highlight the potential of weakly supervised learning as an annotation-efficient strategy for the analysis of complex structures in X-ray tomographic data, and suggest that alternative loss formulations tailored to sparse annotations may further enhance performance.