基于深度学习和形状位置先验的肺动脉形状分割方法  

Deep learning for accurate lung artery segmentation with shape-position priors

作　　者：郭超[1] 高雪涵 胡启迪 李健 朱海星 赵珂 刘卫朋 李单青 GUO Chao;GAO Xuehan;HU Qidi;LI Jian;ZHU Haixing;ZHAO Ke;LIU Weipeng;LI Shanqing(Department of Thoracic Surgery,Peking Union Medical College Hospital,Chinese Academy of Medical Sciences and Peking Union Medical College,Beijing,100730,P.R.China;Hebei University of Technology,School of Artificial Intelligence and Data Science,Tianjin,300401,P.R.China;Hubei Industrial Internet Industry Technology Research Institute,Huangshi,431602,Hubei,P.R.China)

机构地区：[1]中国医学科学院北京协和医学院北京协和医院胸外科,北京100730 [2]河北工业大学人工智能与数据科学学院,天津300401 [3]湖北省工业互联网产业技术研究院,湖北黄石431602

出　　处：《中国胸心血管外科临床杂志》2025年第3期332-338,共7页Chinese Journal of Clinical Thoracic and Cardiovascular Surgery

基　　金：中央高水平医院临床科研业务费(2022-PUMCH-B-011);中国医学科学院临床与转化医学研究专项(2023-I2MC&T-B-019)。

摘　　要：目的提出一种融合形状和位置先验的肺动脉分割方法,旨在解决CT影像下肺动脉与周围组织相似性高、尺寸差异小所导致的分割不精确等问题。方法基于三维U-Net网络架构,依托PARSE 2022数据库影像数据,引入形状和位置先验知识,设计特征提取和融合策略,以增强肺动脉分割能力。将患者数据分为3组:训练集、验证集和测试集。评估模型性能指标包括Dice相似系数(DSC)、灵敏度、精度和豪斯多夫距离(HD95)。结果研究共纳入203例患者的肺动脉影像数据,包括训练集100例、验证集30例和测试集73例。通过主干网络对肺动脉进行粗分割,获得完整的血管结构;利用融合形状和位置信息分支网络提取肺内小动脉特征,减少肺动脉干和左右肺动脉的干扰。实验结果表明,所构建的基于形状和位置先验的分割模型与传统三维U-Net和V-Net方法相比,具有较高的DSC(82.81%±3.20%vs.80.47%±3.17%vs.80.36%±3.43%)、灵敏度(85.30%±8.04%vs.80.95%±6.89%vs.82.82%±7.29%)和精度(81.63%±7.53%vs.81.19%±8.35%vs.79.36%±8.98%)。HD95可达(9.52±4.29)mm,较传统方法短6.05 mm,在分割边界上具有优秀的表现。结论基于形状和位置先验的肺动脉分割方法能够实现肺动脉血管的精确分割,在构建支气管镜或经皮穿刺手术导航任务中具有潜在的应用价值。Objective To propose a lung artery segmentation method that integrates shape and position prior knowledge,aiming to solve the issues of inaccurate segmentation caused by the high similarity and small size differences between the lung arteries and surrounding tissues in CT images.Methods Based on the three-dimensional U-Net network architecture and relying on the PARSE 2022 database image data,shape and position prior knowledge was introduced to design feature extraction and fusion strategies to enhance the ability of lung artery segmentation.The data of the patients were divided into three groups:a training set,a validation set,and a test set.The performance metrics for evaluating the model included Dice Similarity Coefficient(DSC),sensitivity,accuracy,and Hausdorff distance(HD95).Results The study included lung artery imaging data from 203 patients,including 100 patients in the training set,30 patients in the validation set,and 73 patients in the test set.Through the backbone network,a rough segmentation of the lung arteries was performed to obtain a complete vascular structure;the branch network integrating shape and position information was used to extract features of small pulmonary arteries,reducing interference from the pulmonary artery trunk and left and right pulmonary arteries.Experimental results showed that the segmentation model based on shape and position prior knowledge had a higher DSC(82.81%±3.20%vs.80.47%±3.17%vs.80.36%±3.43%),sensitivity(85.30%±8.04%vs.80.95%±6.89%vs.82.82%±7.29%),and accuracy(81.63%±7.53%vs.81.19%±8.35%vs.79.36%±8.98%)compared to traditional three-dimensional U-Net and V-Net methods.HD95 could reach(9.52±4.29)mm,which was 6.05 mm shorter than traditional methods,showing excellent performance in segmentation boundaries.Conclusion The lung artery segmentation method based on shape and position prior knowledge can achieve precise segmentation of lung artery vessels and has potential application value in tasks such as bronchoscopy or percutaneous puncture surgery navigati

关 键 词：肺动脉分割 深度学习 三维U-Net 先验知识 手术导航 

分 类 号：R734.2[医药卫生—肿瘤] TP391.41[医药卫生—临床医学] TP18[自动化与计算机技术—计算机应用技术]