Zhang Peipei, Min Xiangde, Feng Zhaoyan, Kang Zhen, Li Basen, Cai Wei, Fan Chanyuan, Yin Xi, Xie Jinke, Lv Wenzhi, Wang Liang
Department of Radiology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430030, People's Republic of China.
Department of Artificial Intelligence, Julei Technology Company, Wuhan 430030, People's Republic of China.
Cancer Manag Res. 2021 Jan 28;13:839-847. doi: 10.2147/CMAR.S288378. eCollection 2021.
To compare the performance of histogram analysis and intra-perinodular textural transition (Ipris) for distinguishing between benign and malignant testicular lesions.
This retrospective study included 76 patients with 80 pathologically confirmed testicular lesions (55 malignant, 25 benign). All patients underwent preoperative T2-weighted imaging (T2WI) on a 3.0T MR scanner. All testicular lesions were manually segmented on axial T2WI, and histogram and Ipris features were extracted. Thirty enrolled patients were randomly selected to estimate the robustness of the features. We used intraclass correlation coefficients (ICCs) to evaluate intra- and interobserver agreement of features, independent -test or Mann-Whitney -test to compare features between benign and malignant lesions, and receiver operating characteristic curve analysis to evaluate the diagnostic performance of features.
Eighteen histogram features and forty-eight Ipris features were extracted from T2WI of each lesion. Most (60/66) histogram and Ipris features had good robustness (ICC of both intra- and interobserver variabilities >0.6). Three histogram and nine Ipris features were significantly different between the benign and malignant groups. The area under the curve values for Energy, TotalEnergy, and Ipris_shell1_id_std were 0.807, 0.808, and 0.708, respectively, which were relatively higher than those of other features.
Ipris features may be useful for identifying benign and malignant testicular tumors but have no significant advantage over conventional histogram features.
比较直方图分析和结节内纹理过渡（Ipris）在鉴别睾丸良恶性病变中的性能。
本回顾性研究纳入了76例患者，共80个经病理证实的睾丸病变（55个恶性，25个良性）。所有患者均在3.0T MR扫描仪上进行了术前T2加权成像（T2WI）。所有睾丸病变均在轴向T2WI上进行手动分割，并提取直方图和Ipris特征。随机选择30例入组患者评估特征的稳健性。我们使用组内相关系数（ICC）来评估特征在观察者内和观察者间的一致性，使用独立样本t检验或曼-惠特尼U检验来比较良恶性病变之间的特征，并使用受试者工作特征曲线分析来评估特征的诊断性能。
从每个病变的T2WI中提取了18个直方图特征和48个Ipris特征。大多数（60/66）直方图和Ipris特征具有良好的稳健性（观察者内和观察者间变异的ICC均>0.6）。良性和恶性组之间有3个直方图特征和9个Ipris特征存在显著差异。能量、总能量和Ipris_shell1_id_std的曲线下面积值分别为0.807、0.808和0.708，相对高于其他特征。
Ipris特征可能有助于鉴别睾丸良恶性肿瘤，但与传统的直方图特征相比没有显著优势。
