Few-Shot 3D Volumetric Segmentation with Multi-Surrogate Fusion

TL;DR

This paper introduces MSFSeg, a lightweight few-shot 3D segmentation framework that leverages multi-surrogate fusion to accurately segment unseen objects with minimal annotations, reducing training data needs and improving cross-domain generalization.

Contribution

The paper proposes a novel multi-surrogate fusion module for few-shot 3D segmentation, enabling automatic segmentation of unseen objects with limited annotated slices or sequences.

Findings

Outperforms prior methods on few-shot segmentation benchmarks.

Achieves remarkable cross-domain segmentation performance on proprietary datasets.

Requires significantly fewer labeled data for accurate 3D segmentation.

Abstract

Conventional 3D medical image segmentation methods typically require learning heavy 3D networks (e.g., 3D-UNet), as well as large amounts of in-domain data with accurate pixel/voxel-level labels to avoid overfitting. These solutions are thus extremely time- and labor-expensive, but also may easily fail to generalize to unseen objects during training. To alleviate this issue, we present MSFSeg, a novel few-shot 3D segmentation framework with a lightweight multi-surrogate fusion (MSF). MSFSeg is able to automatically segment unseen 3D objects/organs (during training) provided with one or a few annotated 2D slices or 3D sequence segments, via learning dense query-support organ/lesion anatomy correlations across patient populations. Our proposed MSF module mines comprehensive and diversified morphology correlations between unlabeled and the few labeled slices/sequences through multiple designated surrogates, making it able to generate accurate cross-domain 3D segmentation masks given annotated slices or sequences. We demonstrate the effectiveness of our proposed framework by showing superior performance on conventional few-shot segmentation benchmarks compared to prior art, and remarkable cross-domain cross-volume segmentation performance on proprietary 3D segmentation datasets for challenging entities, i.e., tubular structures, with only limited 2D or 3D labels.