Johnson-Lindenstrauss Lemma Guided Network for Efficient 3D Medical Segmentation

2025-09-26

AI Summary

Lightweight 3D medical image segmentation faces a fundamental trade-off between efficiency and robustness, particularly under complex anatomical structures and heterogeneous imaging modalities, where feature representations remain fragile. To address this, we propose VeloxSeg—a novel dual-stream CNN-Transformer collaborative architecture. It integrates Johnson-Lindenstrauss-guided lightweight convolution (JLC), paired-window attention, spatially decoupled knowledge transfer, and Gram-matrix-driven self-supervised knowledge distillation to enable efficient multimodal feature fusion and highly robust representation learning. Evaluated on mainstream 3D multimodal benchmarks, VeloxSeg achieves a 26% improvement in Dice coefficient over prior lightweight methods. It attains 11× higher GPU throughput and 48× faster CPU inference speed compared to state-of-the-art lightweight models, significantly breaking through the performance bottleneck of lightweight segmentation architectures.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data

Abstract

Lightweight 3D medical image segmentation remains constrained by a fundamental "efficiency / robustness conflict", particularly when processing complex anatomical structures and heterogeneous modalities. In this paper, we study how to redesign the framework based on the characteristics of high-dimensional 3D images, and explore data synergy to overcome the fragile representation of lightweight methods. Our approach, VeloxSeg, begins with a deployable and extensible dual-stream CNN-Transformer architecture composed of Paired Window Attention (PWA) and Johnson-Lindenstrauss lemma-guided convolution (JLC). For each 3D image, we invoke a "glance-and-focus" principle, where PWA rapidly retrieves multi-scale information, and JLC ensures robust local feature extraction with minimal parameters, significantly enhancing the model's ability to operate with low computational budget. Followed by an extension of the dual-stream architecture that incorporates modal interaction into the multi-scale image-retrieval process, VeloxSeg efficiently models heterogeneous modalities. Finally, Spatially Decoupled Knowledge Transfer (SDKT) via Gram matrices injects the texture prior extracted by a self-supervised network into the segmentation network, yielding stronger representations than baselines at no extra inference cost. Experimental results on multimodal benchmarks show that VeloxSeg achieves a 26% Dice improvement, alongside increasing GPU throughput by 11x and CPU by 48x. Codes are available at https://github.com/JinPLu/VeloxSeg.

Problem

• Addresses lightweight 3D medical segmentation efficiency-robustness conflict
• Redesigns framework using high-dimensional 3D image characteristics
• Enhances multimodal modeling with minimal computational parameters

Innovation

• Dual-stream CNN-Transformer with PWA and JLC modules
• Modal interaction in multi-scale image-retrieval process
• Spatially Decoupled Knowledge Transfer via Gram matrices