Shape Features of the Lesion Habitat to Differentiate Brain Tumor Progression from Pseudoprogression on Routine Multiparametric MRI: A Multisite Study.
Shape Features of the Lesion Habitat to Differentiate Brain Tumor Progression from Pseudoprogression on Routine Multiparametric MRI: A Multisite Study.
复制标题
DOI：
10.3174/ajnr.a5858
复制
发表时间：
2018-12
期刊：
影响因子：
--
通讯作者：
Tiwari P
中科院分区：
文献类型：
--
作者：
Ismail M;Hill V;Statsevych V;Huang R;Prasanna P;Correa R;Singh G;Bera K;Beig N;Thawani R;Madabhushi A;Aahluwalia M;Tiwari P
区分假进展（PsP），一种放射诱导的治疗效应，与成像上的肿瘤进展是胶质母细胞瘤管理中的一个重大挑战。不幸的是，RANO标准设定的指南仅基于T1 w、T2 w/FLAIR扫描上观察到的增强的双向直径测量。我们假设T1 w和T2 w/FLAIR高信号增强病变的定量3D形状特征（统称为病变栖息地）可以更全面地捕获PsP和肿瘤复发的病理生理差异，而仅凭直径测量是不可感知的。共分析了来自2家机构的105项胶质母细胞瘤研究，包括培训（N=59）和独立测试（N=46）队列。对于每项研究，获得病变栖息地（T1 w增强病变和T2 w/FLAIR高信号病变周围区域）的专家描绘，然后提取30个形状特征，捕获14个“全局”轮廓特征和16个“局部”曲率测量，每个栖息地区域。采用特征选择来识别训练组群上的最有区别的特征，并使用支持向量机分类器对测试组群进行评估。前2个最具鉴别力的特征被确定为捕获增强病变的总曲率的局部特征和T2 w/FLAIR高信号病变周围区域的曲率。使用来自训练队列的最佳特征（训练准确率=91.5%），我们在区分PsP与肿瘤进展的测试集上获得了90.2%的准确率。我们的初步结果表明，病变栖息地的3D形状属性可以在PsP和肿瘤进展中差异表达，并可用于区分这些放射学相似的病理。
Differentiating pseudo-progression (PsP), a radiation-induced treatment effect, from tumor progression on imaging is a significant challenge in Glioblastoma management. Unfortunately, guidelines set by RANO criteria are based solely on bidirectional diametric measurements of enhancement observed on T1w, T2w/FLAIR scans. We hypothesize that quantitative 3D shape features of the enhancing lesion on T1w, and T2w/FLAIR hyperintensities (together called the lesion habitat) can more comprehensively capture pathophysiological differences across PsP and tumor recurrence, not appreciable on diametric measurements alone. A total of 105 Glioblastoma studies from 2 institutions were analyzed, consisting of a training (N=59) and an independent test (N=46) cohort. For every study, expert delineation of the lesion habitat (T1w enhancing lesion and T2w/FLAIR hyperintense peri-lesional region) was obtained, followed by extracting 30 shape features capturing 14 “global” contour characteristics, and 16 “local” curvature measures, for every habitat region. Feature selection was employed to identify most discriminative features on the training cohort and were evaluated on the test cohort using a support vector machine classifier. Top 2 most discriminative features were identified as local features capturing total curvature of the enhancing lesion, and curvedness of T2w/FLAIR hyperintense peri-lesional region. Using top features from the training cohort (training accuracy=91.5%), we obtained an accuracy of 90.2% on the test set in distinguishing PsP from tumor progression. Our preliminary results suggest that 3D shape attributes from the lesion habitat can differentially express across PsP and tumor progression and could be used to distinguish these radiographically-similar pathologies.