基于扩散峰度成像序列平均扩散峰度图的影像组学列线图在鉴别乳腺良恶性肿块中的应用  

Application of radiomics nomograms based on DKI sequence MK map in differentiating benign and malignant breast masses

作　　者：曹蕾 周静怡 蒋璟璇 CAO Lei;ZHOU Jing-yi;JIANG Jing-xuan(Department of Radiology,Nantong Haimen District Traditional Chinese Medicine Hospital,Nantong 226100,Jiangsu,CHINA;Department of Radiology,the Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University,Shanghai 200233,CHINA;Department of Radiology,the Second People’s Hospital of Kunshan,Kunshan 215300,Jiangsu,CHINA)

机构地区：[1]南通市海门区中医院放射科,江苏南通226100 [2]上海交通大学附属第六人民医院放射科,上海200233 [3]昆山市第二人民医院放射科,江苏昆山215300

出　　处：《海南医学》2024年第13期1909-1913,共5页Hainan Medical Journal

摘　　要：目的探讨基于磁共振扩散峰度成像(DKI)序列平均扩散峰度(MK)图像影像组学特征构建列线图模型术前预测乳腺肿块良恶性的价值。方法纳入2018年1月至2022年12月于上海交通大学附属第六人民医院术前行DKI检查的乳腺肿块患者129例。根据术后病理分为良性组55例和恶性组74例,按照7:3随机分为训练集87例和验证集42例。基于MK图像提取肿块的影像组学特征,构建影像组学模型。使用逻辑回归(LR)算法结合影像组学特征和独立临床危险因素构建列线图模型,并绘制受试者特征曲线(ROC)以评估各模型的诊断效能。结果最后共筛出3个影像组学特征和2个临床特征来构建模型。在测试队列中,列线图模型曲线下面积(AUC)为0.947,明显高于影像组学模型(AUC:0.854)和临床模型(AUC:0.789),显示出最佳效能(均P<0.05)。结论联合影像组学和临床特征构建的列线图模型能够更好地在术前鉴别乳腺肿块的良恶性。Objective To explore the value of radiomics nomograms based on mean kurtosis(MK)map of diffusion kurtosis imaging(DKI)sequence for preoperatively predicting the benign or malignant breast masses.Methods This study included 129 patients with breast masses who underwent DKI at the the Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University from January 2018 to December 2022.Patients were categorized into benign(55 cases)and malignant groups(74 cases)based on postoperative pathology,and then randomly divided into a training set(87 cases)and a validation set(42 cases)at a ratio of 7:3.Radiomics of the tumors were extracted from the MK map to construct a radiomics model.A logistic regression(LR)algorithm was used in combination with radiomics features and independent clinical risk factors to construct a nomogram model,and receiver operating characteristic(ROC)curves were plotted to evaluate the diagnostic efficacy of each model.Results A total of 3 radiomics features and 2 clinical features were selected to construct the models.In the validation cohort,the area under the curve(AUC)of the nomogram model was 0.947,which was significantly higher than that of the radiomics model(AUC:0.854)and the clinical model(AUC:0.789),demonstrating the best performance(all P<0.05).Conclusion A nomogram model constructed by integrating radiomics and clinical features can better differentiate the benign and malignant nature of breast masses preoperatively.

关 键 词：扩散峰度成像 磁共振 影像组学 乳腺 列线图 

分 类 号：R737.9[医药卫生—肿瘤]