Hybrid graph convolutional neural networks for landmark-based anatomical segmentation

TL;DR

HybridGNet is a novel neural architecture combining image and graph convolutions for landmark-based anatomical segmentation, demonstrating robustness to occlusions and producing plausible, shape-constrained segmentations.

Contribution

It introduces HybridGNet, integrating standard and graph convolutions, to improve landmark-based segmentation accuracy and robustness in medical imaging.

Findings

HybridGNet outperforms standard models in robustness to occlusions.

It can generate landmark-based segmentations from pixel annotations.

The model produces anatomically plausible results with shape constraints.

Abstract

In this work we address the problem of landmark-based segmentation for anatomical structures. We propose HybridGNet, an encoder-decoder neural architecture which combines standard convolutions for image feature encoding, with graph convolutional neural networks to decode plausible representations of anatomical structures. We benchmark the proposed architecture considering other standard landmark and pixel-based models for anatomical segmentation in chest x-ray images, and found that HybridGNet is more robust to image occlusions. We also show that it can be used to construct landmark-based segmentations from pixel level annotations. Our experimental results suggest that HybridGNet produces accurate and anatomically plausible landmark-based segmentations, by naturally incorporating shape constraints within the decoding process via spectral convolutions.