Implicit Shape-Prior for Few-Shot Assisted 3D Segmentation

2025-09-10

AI Summary

Manual annotation of medical 3D images is prohibitively costly, and existing segmentation methods struggle to generalize to sparse annotations or previously unseen anatomical structures. Method: We propose a few-shot segmentation framework integrating implicit neural representations with learnable shape priors. Specifically, we design a parametric implicit shape prior model capable of reconstructing full 3D anatomy from as few as one to three interactively annotated 2D slices; incorporate an active learning strategy to intelligently select the most informative slices, minimizing clinician interaction; and enable zero-shot cross-organ and unseen-muscle generalization. Contribution/Results: Evaluated on two challenging tasks—organs-at-risk segmentation in brain tumor MRI and a novel sarcopenia dataset—the framework achieves state-of-the-art accuracy with minimal slice-level annotations, improving annotation efficiency by 5–8× and advancing clinical deployment of few-shot assisted segmentation.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Active Learning
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Semantics and Knowledge: Representation, semantic annotation, enhancement, enrichments, access and/or integration of a variety of data on the Web

Abstract

The objective of this paper is to significantly reduce the manual workload required from medical professionals in complex 3D segmentation tasks that cannot be yet fully automated. For instance, in radiotherapy planning, organs at risk must be accurately identified in computed tomography (CT) or magnetic resonance imaging (MRI) scans to ensure they are spared from harmful radiation. Similarly, diagnosing age-related degenerative diseases such as sarcopenia, which involve progressive muscle volume loss and strength, is commonly based on muscular mass measurements often obtained from manual segmentation of medical volumes. To alleviate the manual-segmentation burden, this paper introduces an implicit shape prior to segment volumes from sparse slice manual annotations generalized to the multi-organ case, along with a simple framework for automatically selecting the most informative slices to guide and minimize the next interactions. The experimental validation shows the method's effectiveness on two medical use cases: assisted segmentation in the context of at risks organs for brain cancer patients, and acceleration of the creation of a new database with unseen muscle shapes for patients with sarcopenia.

Problem

• Reducing manual workload in 3D medical segmentation
• Segmenting organs at risk from sparse annotations
• Automating slice selection for efficient muscle volume analysis

Innovation

• Implicit shape prior for segmentation
• Automatic selection of informative slices
• Generalized to multi-organ medical cases