LGMSNet: Thinning a medical image segmentation model via dual-level multiscale fusion

2025-08-21

AI Summary

Existing lightweight medical image segmentation models often sacrifice attention mechanisms, leading to insufficient global contextual modeling, and neglect channel redundancy within convolutional kernels at the same layer, thereby limiting feature representation and generalization. To address these issues, we propose a dual-level multi-scale fusion framework: (1) heterogeneous intra-layer convolutional kernels to mitigate channel redundancy; (2) a sparse Transformer-convolution hybrid branch for efficient capture of global low-frequency context; and (3) local-global collaborative fusion to enhance multi-scale feature representation. The method maintains model lightness while significantly improving contextual awareness and zero-shot transferability. It outperforms state-of-the-art methods on six benchmark datasets and achieves superior zero-shot segmentation performance on four unseen datasets, demonstrating strong potential for clinical deployment.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multi-instance/Multi-view 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
• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Medical image segmentation plays a pivotal role in disease diagnosis and treatment planning, particularly in resource-constrained clinical settings where lightweight and generalizable models are urgently needed. However, existing lightweight models often compromise performance for efficiency and rarely adopt computationally expensive attention mechanisms, severely restricting their global contextual perception capabilities. Additionally, current architectures neglect the channel redundancy issue under the same convolutional kernels in medical imaging, which hinders effective feature extraction. To address these challenges, we propose LGMSNet, a novel lightweight framework based on local and global dual multiscale that achieves state-of-the-art performance with minimal computational overhead. LGMSNet employs heterogeneous intra-layer kernels to extract local high-frequency information while mitigating channel redundancy. In addition, the model integrates sparse transformer-convolutional hybrid branches to capture low-frequency global information. Extensive experiments across six public datasets demonstrate LGMSNet's superiority over existing state-of-the-art methods. In particular, LGMSNet maintains exceptional performance in zero-shot generalization tests on four unseen datasets, underscoring its potential for real-world deployment in resource-limited medical scenarios. The whole project code is in https://github.com/cq-dong/LGMSNet.

Problem

• Lightweight medical image segmentation with minimal performance compromise
• Addressing channel redundancy and global context limitations in models
• Enhancing feature extraction for resource-constrained clinical settings

Innovation

• Heterogeneous kernels reduce channel redundancy
• Sparse transformer-convolutional hybrid branches
• Dual-level multiscale fusion for lightweight segmentation