| 田朋,李金锋,李晶,陈穗惠,王新江,徐贤.MRI影像组学鉴别前列腺影像报告和数据系统(PI-RADS)3分良、恶性病变[J].中国医学影像技术,2024,40(12):1920~1925 |
| MRI影像组学鉴别前列腺影像报告和数据系统(PI-RADS)3分良、恶性病变 |
| MRI radiomic for differentiating benign and malignant lesions with prostate imaging reporting and data system (PI-RADS) 3 points |
| 投稿时间：2024-09-29 修订日期：2024-11-03 |
| DOI：10.13929/j.issn.1003-3289.2024.12.024 |
| 中文关键词: 前列腺肿瘤 磁共振成像 影像组学 |
| 英文关键词:prostatic neoplasms magnetic resonance imaging radiomics |
| 基金项目: |
|
| 摘要点击次数: 361 |
| 全文下载次数: 91 |
| 中文摘要: |
| 目的 评价MRI影像组学鉴别前列腺影像报告和数据系统(PI-RADS)3分良、恶性病变的价值。方法 回顾性分析107例PI-RADS 3分前列腺病变患者,按73比例随机将其分为训练集与测试集。采用回归分析筛选与前列腺癌(PCa)相关的临床特征,以最小绝对收缩和选择算子算法筛选病变影像组学特征,分别构建临床模型、单序列模型、多序列模型及多序列-临床联合模型。绘制受试者工作特征曲线,计算曲线下面积(AUC),评估各模型鉴别良、恶性病变的效能;并以校准曲线和决策曲线评价模型性能及临床实用性。结果 总前列腺特异性抗原(PSA)为PCa独立临床预测因素,以之构建临床模型;分别以4个表观弥散系数(ADC)特征、10个弥散加权成像(DWI)特征及14个T2WI特征构建单序列模型;利用2个ADC特征、3个DWI特征及3个T2WI特征构建多序列组学模型。以所获联合模型鉴别训练集与测试集前列腺PI-RADS 3分良、恶性病变的AUC分别为0.940和0.906,提示其具有良好预测效能和临床净获益。结论 基于MR多序列影像组学特征联合总PSA可鉴别PI-RADS 3分前列腺良、恶性病变。 |
| 英文摘要: |
| Objective To investigate the value of MRI radiomics for differentiating benign and malignant prostate lesions with prostate imaging reporting and data system (PI-RADS) 3 points. Methods Data of 107 patients with PI-RADS scoring of 3 points lesions were retrospectively analyzed. The patients were randomly divided into training set and testing set at a 7∶3 ratio. Clinical factors related to prostate cancer (PCa) were screened with regression analysis, and lesions' radiomics features were extracted with the least absolute shrinkage and selection operator algorithm, and then clinical models, single sequence models, multiple sequence model and combined model (multiple sequence+clinical) were constructed, respectively. Then receiver operating characteristic curves were drawn, and the area under the curve (AUC) was calculated, the efficacy of each model for differentiating PI-RADS 3 points benign and malignant lesions were assessed, and their performance and clinical applicability were evaluated using the calibration curve and clinical decision curve. Results The total prostate-specific antigen (PSA) was the clinical factor related to PCa, based on which a clinical model was established. Four apparent diffusion coefficient (ADC) features, 10 diffusion-weighted imaging (DWI) features and 14 T2WI features of lesions were used to construct single sequence models, while 2 ADC features, 3 DWI features and 3 T2WI features of lesions were selected to construct multiple sequence model. AUC of the combined model for differentiating benign and malignant PI-RADS 3 points lesions in training and test sets was 0.940 and 0.906, respectively, indicating that the combined model had good predictive performance and clinical net benefit. Conclusion The combined model established based on multiple sequence radiomics features and total PSA was helpful to differentiating PI-RADS 3 points benign and malignant prostate lesions. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
