SSAT-Swin:基于自监督Swin Transformer的深度学习脊柱超声特征分割用于脊柱侧凸
本研究提出基于Transformer的多类分割框架SSAT-Swin,用于脊柱侧凸超声图像分析,在解码器集成边界增强模块、跳跃连接集成通道注意力模块,并在1,170张图像上自监督预训练后于109对图像-标签上微调。模型Dice评分85.6%、Jaccard评分74.5%、脊柱侧凸骨特征检出率92.8%,优于当前先进模型。
要点
- SSAT-Swin在109对图像-标签微调后Dice评分85.6%、Jaccard评分74.5%。
- 脊柱侧凸骨特征检出率达92.8%,优于所比较的先进模型。
- 自监督预训练使用1,170张图像,提升模型捕获全局上下文信息的能力。
核心结论
- 自监督学习增强模型捕获全局上下文信息的能力,使其适合应对超声图像的独特挑战,通过更准确的分割推进脊柱侧凸评估。
研究设计与样本
- 基于Transformer的多类分割框架SSAT-Swin,在1,170张图像上进行自监督代理任务预训练,随后在109对图像-标签上微调;解码器集成边界增强模块,跳跃连接集成通道注意力模块。
诊断标准与关键数值
- SSAT-Swin达到Dice评分85.6%、Jaccard评分74.5%、脊柱侧凸骨特征检出率92.8%,优于当前先进模型。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
OBJECTIVE: Scoliosis, a 3-D spinal deformity, requires early detection and intervention. Ultrasound curve angle (UCA) measurement using ultrasound images has emerged as a promising diagnostic tool. However, calculating the UCA directly from ultrasound images remains challenging due to low contrast, high noise, and irregular target shapes. Accurate segmentation results are therefore crucial to enhance image clarity and precision prior to UCA calculation.
METHODS: We propose the SSAT-Swin model, a transformer-based multi-class segmentation framework designed for ultrasound image analysis in scoliosis diagnosis. The model integrates a boundary-enhancement module in the decoder and a channel attention module in the skip connections. Additionally, self-supervised proxy tasks are used during pre-training on 1,170 images, followed by fine-tuning on 109 image-label pairs.
RESULTS: The SSAT-Swin achieved Dice scores of 85.6% and Jaccard scores of 74.5%, with a 92.8% scoliosis bone feature detection rate, outperforming state-of-the-art models.
CONCLUSION: Self-supervised learning enhances the model's ability to capture global context information, making it well-suited for addressing the unique challenges of ultrasound images, ultimately advancing scoliosis assessment through more accurate segmentation.
引用
- Zhang C, Zheng Y, McAviney J, et al. SSAT-Swin: Deep Learning-Based Spinal Ultrasound Feature Segmentation for Scoliosis Using Self-Supervised Swin Transformer. Ultrasound Med Biol. 2025. PMID: 40082183. doi:10.1016/j.ultrasmedbio.2025.02.013
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。