本文已被:浏览 523次 下载 201次 |
码上扫一扫! |
|
基于MRI影像组学机器学习模型鉴别小肾癌与乏脂肪肾血管平滑肌脂肪瘤 |
王睿婷1,2, 钟莲婷3, 潘先攀4, 陈磊4, 曾蒙苏1,2, 丁玉芹1,2, 周建军3,5,6
|
1.上海市影像医学研究所, 上海 200032;2.复旦大学附属中山医院放射科, 上海 200032;3.复旦大学附属中山医院厦门医院放射科, 厦门 361015;4.上海联影智能有限公司, 上海 200232;5.厦门市影像医学临床医学研究中心, 厦门 361015;6.厦门市放射科临床重点专科, 厦门 361015
|
|
摘要: |
目的 探讨基于多期相MRI影像组学机器学习模型鉴别小肾癌(small renal cell carcinoma,sRCC)与乏脂肪肾血管平滑肌脂肪瘤(fat-poor renal angiomyolipoma,fp-AML)的价值。方法 回顾性分析经病理证实的79例sRCCs与35例最大径≤4 cm的fp-AMLs。分别在T2WI(T2),增强前(UP)、皮髓质期(CMP)、实质期(NP)T1WI图像上手工勾画全肿瘤感兴趣区(volume of interest,VOI),并提取影像组学特征。以7∶3划分训练集(n=86)与测试集(n=28),采用t检验、最大相关最小冗余(maximal relevance and minimal redundancy,mRMR)算法、最小绝对收缩和选择算子方法(the least absolute shrinkage and selection operator,LASSO)进行特征选择,分别建立逻辑回归(logistic regression,LR)与支持向量机(support vector machine,SVM)分类模型。采用受试者工作特征(ROC)曲线来评价模型的分类性能。结果 分别从T2、UP、CMP、NP和4期相整合图像中获得4、12、3、11、15个最优子集特征。基于NP或4期相整合影像组学特征联合LR构建的模型鉴别效能最佳;模型在训练集中的AUC分别为0.956、0.986,在测试集中的AUC均为0.881。结论 基于多期相MRI影像组学特征构建的机器学习模型在鉴别sRCC和fp-AML中具有较高的效能。 |
关键词: 磁共振成像 影像组学 小肾癌 乏脂肪肾血管平滑肌脂肪瘤 |
DOI:10.12025/j.issn.1008-6358.2023.20222164 |
分类号: |
基金项目:福建省科技计划项目引导性项目(2019D025),福建省卫生健康科研人才培养项目医学创新课题(2019CXB33). |
|
MRI-based radiomics machine learning model for differentiating small renal cell carcinoma from fat-poor renal angiomyolipoma |
WANG Rui-ting1,2, ZHONG Lian-ting3, PAN Xian-pan4, CHEN Lei4, ZENG Meng-su1,2, DING Yu-qin1,2, ZHOU Jian-jun3,5,6
|
1.Shanghai Institute of Medical Imaging, Shanghai 200032, China;2.Department of Radiology, Zhongshan Hospital, Fudan University, Shanghai 200032, China;3.Department of Radiology, Zhongshan Hospital (Xiamen Branch), Fudan University, Xiamen 361015, Fujian, China;4.Shanghai United Imaging Intelligence Co., Ltd. Shanghai, Shanghai 200232, China;5.Xiamen Municipal Clinical Research Center for Medical Imaging, Xiamen 361015, Fujian, China;6.Xiamen Key Clinical Specialty for Radiology, Xiamen 361015, Fujian, China
|
Abstract: |
Objective To investigate the value of multi-phase MRI-based radiomics machine learning models in differentiating small renal cell carcinoma (sRCC) from fat-poor renal angiomyolipoma (fp-AML). Methods 79 cases of sRCCs and 35 cases of fp-AMLs (diameter ≤ 4 cm) which were confirmed by pathology were retrospectively analyzed. The volume of interest (VOI) of the total tumor was manually delineated on the images of T2WI (T2), unenhanced phase (UP), corticomedullary phase (CMP) and nephrographic phase (NP) and then the radiomics of the VOIs were extracted respectively. The training set and the test set were set according to the ratio of 7:3. The t-test, maximal relevance and minimal redundancy (mRMR) and the least absolute shrinkage and selection operator (LASSO) were used to select the radiomics features. The selected features were used to build classification models with logistic regression (LR) and support vector machine (SVM). The receiver operating characteristic (ROC) curve was used to evaluate the classification performances of the models. Results There were 4, 12, 3, 11 and 15 optimal features obtained from T2、UP、CMP、NP and the combined four phases, respectively. The radiomics features based on NP or the combined four phases with LR model performed best, AUCs were respectively 0.956, 0.986 in the training set and both were 0.881 in the test set. Conclusion The multi-phase MRI-based radiomics machine learning model has favorable diagnostic performance in differentiating sRCC from fp-AML. |
Key words: magnetic resonance imaging radiomics small renal cell carcinoma fat-poor renal angiomyolipoma |