基于注意力机制的腰椎间盘突出患者多裂肌分割方法

吉林大学学报(信息科学版) ›› 2023, Vol. 41 ›› Issue (5): 876-884.

基于注意力机制的腰椎间盘突出患者多裂肌分割方法

李夏¹ , 胡巍² , 王子民¹ , 贺泽华¹ , 周悦¹ , 关挺强¹ , 郭欣¹

1. 桂林电子科技大学计算机与信息安全学院, 广西桂林 541000; 2. 桂林市人民医院脊柱骨病外科, 广西桂林 541000

收稿日期:2022-10-20 出版日期:2023-10-09 发布日期:2023-10-10
通讯作者: 王子民(1974— ), 男, 山东聊城人, 桂林电子科技大学副教授, 主要从事智能信息处理、模式识别等研究, (Tel)86-18877388999(E-mail)worthyman@ guet. edu. cn。 E-mail:worthyman@ guet. edu. cn
作者简介:李夏(1998— ), 女, 四川德阳人, 桂林电子科技大学硕士研究生, 主要从事图像处理研究, ( Tel) 86-13036591531 (E-mail)LIXia_Lee12@ 163. com
基金资助:
国家自然科学基金资助项目(61866009); 广西重点研发计划基金资助项目(AB21220037); 桂林市科学研究与技术开发基金资助项目(20210227-2); 桂林电子科技大学研究生教育创新基金资助项目(2020YCXS048)

Segmentation of Multifidus Muscle in Patients with Lumbar Disc Herniation Based on Attention Mechanism

LI Xia ¹ , HU Wei ² , WANG Zimin ¹ , HE Zehua ¹ , ZHOU Yue ¹ , GUAN Tingqiang ¹ , GUO Xin ¹

1. School of Computer Science and Information Security, Guilin University of Electronic Technology, Guilin 541000, China; 2. Department of Spine and Osteopathy Ward, Guilin Peoples’ Hospital, Guilin 541000, China

Received:2022-10-20 Online:2023-10-09 Published:2023-10-10
Contact: E-mail:worthyman@ guet. edu. cn

摘要/Abstract

摘要： 为解决分割目标和周围结构边界不清楚的问题, 提出一种基于注意力机制的腰椎间盘突出患者多裂肌分割方法。该网络采用了编码器-解码器的结构, 通过引入注意力机制模块提升网络分割精度, 并在特征提取后引入空洞空间卷积池化金字塔模块, 融合了上下文信息, 以提升网络模型的性能。实验结果表明, 在推理时间接近的情况下, 该模型与经典 U-Net 算法相比, Dice 系数提升了 7. 8% , Jaccard 相似系数提升了10. 1% , Hausdorff Distance 下降了 69. 5% , 提高了多裂肌脂肪浸润部位的分割精度。

关键词: 腰椎间盘突出症, 核磁共振成像, U-Net 算法, 注意力机制, 空洞空间卷积池化金字塔

Abstract: Automatic analysis of lumbar disc herniation requires precise segmentation of the multifidus muscle’s fatty infiltration site in spinal MRI ( Magnetic Resonance Imaging) images. An attention-based approach for segmenting the multifidus muscle in lumbar disc herniation patients is proposed to address issues including ambiguous boundaries between segmentation targets and adjacent components. The network utilizes an encoder- decoder structure, and the addition of an attention mechanism module to increase the network segmentation accuracy. After feature extraction, an atrous spatial pyramid pooling module is added to combine contextual data improving the performance of the network model. In comparison to the traditional U-Net algorithm, the experimental results demonstrate that this model improves the segmentation accuracy of the fatty infiltrated regions of multifidus muscle by improving the Dice coefficient by 7. 8% , Jaccard similarity coefficient by 10. 1% , and Hausdorff Distance by 69. 5% .

Key words: lumbar disc herniation(LDH), magnetic resonance imaging(MRI), U-Net algorithm, attention mechanism, atrous spatial pyramid pooling(ASPP)

中图分类号:

TP39

李夏, 胡巍, 王子民, 贺泽华, 周悦, 关挺强, 郭欣. 基于注意力机制的腰椎间盘突出患者多裂肌分割方法[J]. 吉林大学学报(信息科学版), 2023, 41(5): 876-884.

LI Xia , HU Wei , WANG Zimin , HE Zehua , ZHOU Yue , GUAN Tingqiang , GUO Xin .

Segmentation of Multifidus Muscle in Patients with Lumbar Disc Herniation Based on Attention Mechanism [J]. Journal of Jilin University (Information Science Edition), 2023, 41(5): 876-884.