UltraLBM-UNet: Ultralight Bidirectional Mamba-based Model for Skin Lesion Segmentation

2025-12-25

AI Summary

Skin lesion segmentation in primary care faces bottlenecks including low accuracy, high computational overhead, and poor deployability. To address these challenges, we propose UltraLBM-UNet—a lightweight U-Net architecture featuring a novel synergistic mechanism that integrates bidirectional Mamba state-space modeling with multi-branch local feature perception for efficient global-local feature fusion. Furthermore, we introduce a hybrid knowledge distillation strategy to yield an ultra-compact student model, UltraLBM-UNet-T, with only 0.011M parameters and 0.019 GFLOPs. Evaluated on ISIC 2017, ISIC 2018, and PH2 benchmarks, UltraLBM-UNet achieves state-of-the-art segmentation performance while maintaining just 0.034M parameters and 0.060 GFLOPs—significantly outperforming existing lightweight methods. The model enables real-time inference on edge devices, establishing a new paradigm for intelligent, resource-efficient dermatological assistance in low-resource clinical settings.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Skin lesion segmentation is a crucial step in dermatology for guiding clinical decision-making. However, existing methods for accurate, robust, and resource-efficient lesion analysis have limitations, including low performance and high computational complexity. To address these limitations, we propose UltraLBM-UNet, a lightweight U-Net variant that integrates a bidirectional Mamba-based global modeling mechanism with multi-branch local feature perception. The proposed architecture integrates efficient local feature injection with bidirectional state-space modeling, enabling richer contextual interaction across spatial dimensions while maintaining computational compactness suitable for point-of-care deployment. Extensive experiments on the ISIC 2017, ISIC 2018, and PH2 datasets demonstrate that our model consistently achieves state-of-the-art segmentation accuracy, outperforming existing lightweight and Mamba counterparts with only 0.034M parameters and 0.060 GFLOPs. In addition, we introduce a hybrid knowledge distillation strategy to train an ultra-compact student model, where the distilled variant UltraLBM-UNet-T, with only 0.011M parameters and 0.019 GFLOPs, achieves competitive segmentation performance. These results highlight the suitability of UltraLBM-UNet for point-of-care deployment, where accurate and robust lesion analyses are essential. The source code is publicly available at https://github.com/LinLinLin-X/UltraLBM-UNet.

Problem

• Develops a lightweight model for skin lesion segmentation
• Reduces computational complexity while maintaining high accuracy
• Enables point-of-care deployment with minimal parameters and FLOPs

Innovation

• Lightweight U-Net with bidirectional Mamba global modeling
• Multi-branch local feature perception for contextual interaction
• Hybrid knowledge distillation for ultra-compact student model