轻量级医学图像分割模型，效率提升2.5倍且精度更高。
SegMate: Asymmetric Attention-Based Lightweight Architecture for Efficient Multi-Organ Segmentation
- 采用非对称结构与注意力机制，结合多尺度融合与切片位置编码。
- 计算量降低2.5倍、显存减少2.1倍，总分割器上达93.51%骰子分数。
- 适用于资源受限临床场景，跨数据集泛化能力强，适合部署落地。
当前最先进的医学图像分割模型虽精度高，但计算开销大，难以在资源受限的临床环境中部署。本文提出SegMate，一种高效的2.5D框架，在保持顶尖精度的同时显著降低计算需求。其设计融合了非对称架构、注意力机制、多尺度特征融合、基于切片的位置条件化及多任务优化。我们在三种现代骨干网络（EfficientNetV2-M、MambaOut-Tiny、FastViT-T12）上验证了该框架的效率-精度权衡。实验基于TotalSegmentator、SegTHOR和AMOS22三个数据集展开。相比原始模型，SegMate将计算量（GFLOPs）降低最多2.5倍，显存占用（VRAM）减少最多2.1倍，同时性能普遍提升约1%。在TotalSegmentator上，仅用295MB峰值GPU内存即达到93.51%的骰子分数。在SegTHOR和AMOS22上的零样本跨数据集评估显示强泛化能力，骰子分数分别高达86.85%和89.35%。代码已开源：https://github.com/andreibunea99/SegMate。
原文摘要 · Abstract (English)
State-of-the-art models for medical image segmentation achieve excellent accuracy but require substantial computational resources, limiting deployment in resource-constrained clinical settings. We present SegMate, an efficient 2.5D framework that achieves state-of-the-art accuracy, while considerably reducing computational requirements. Our efficient design is the result of meticulously integrating asymmetric architectures, attention mechanisms, multi-scale feature fusion, slice-based positional conditioning, and multi-task optimization. We demonstrate the efficiency-accuracy trade-off of our framework across three modern backbones (EfficientNetV2-M, MambaOut-Tiny, FastViT-T12). We perform experiments on three datasets: TotalSegmentator, SegTHOR and AMOS22. Compared with the vanilla models, SegMate reduces computation (GFLOPs) by up to 2.5x and memory footprint (VRAM) by up to 2.1x, while generally registering performance gains of around 1%. On TotalSegmentator, we achieve a Dice score of 93.51% with only 295MB peak GPU memory. Zero-shot cross-dataset evaluations on SegTHOR and AMOS22 demonstrate strong generalization, with Dice scores of up to 86.85% and 89.35%, respectively. We release our open-source code at https://github.com/andreibunea99/SegMate.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。