Scalable Segmentation for Ultra-High-Resolution Brain MR Images

2025-05-27

AI Summary

To address the challenges of scarce fine-grained annotations, high computational cost, and poor generalization in ultra-high-resolution brain MRI segmentation, this paper proposes a novel class-conditional one-class segmentation paradigm. It leverages easily obtainable low-resolution coarse labels as spatial guidance to regress class-level signed distance transforms (SDTs) instead of discrete segmentation maps, enabling boundary-aware supervision. A lightweight 3D progressive network architecture is designed to support zero-shot transfer to unseen anatomical structures. The method eliminates reliance on expensive high-resolution annotations and alleviates GPU memory bottlenecks. Evaluated on both synthetic and real ultra-high-definition datasets, it significantly outperforms state-of-the-art approaches—reducing training and inference memory consumption by 42%—while enabling plug-and-play generalization to newly introduced classes without retraining.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Although deep learning has shown great success in 3D brain MRI segmentation, achieving accurate and efficient segmentation of ultra-high-resolution brain images remains challenging due to the lack of labeled training data for fine-scale anatomical structures and high computational demands. In this work, we propose a novel framework that leverages easily accessible, low-resolution coarse labels as spatial references and guidance, without incurring additional annotation cost. Instead of directly predicting discrete segmentation maps, our approach regresses per-class signed distance transform maps, enabling smooth, boundary-aware supervision. Furthermore, to enhance scalability, generalizability, and efficiency, we introduce a scalable class-conditional segmentation strategy, where the model learns to segment one class at a time conditioned on a class-specific input. This novel design not only reduces memory consumption during both training and testing, but also allows the model to generalize to unseen anatomical classes. We validate our method through comprehensive experiments on both synthetic and real-world datasets, demonstrating its superior performance and scalability compared to conventional segmentation approaches.

Problem

• Lack of labeled data for ultra-high-resolution brain MRI segmentation
• High computational demands in 3D brain image segmentation
• Difficulty in segmenting fine-scale anatomical structures accurately

Innovation

• Leverages low-resolution labels as spatial guidance
• Regresses signed distance transform maps for boundary-awareness
• Introduces scalable class-conditional segmentation strategy