Department of Mechanical Engineering, Tongji University School of Mechanical Engineering, Shanghai, People's Republic of China. Department of Industrial Engineering, University of Arkansas College of Engineering, Fayetteville, AR, United States of America. Department of Radiation Oncology, University of Washington School of Medicine, Seattle, WA, United States of America.
Phys Med Biol. 2020 Oct 7;65(20):205007. doi: 10.1088/1361-6560/abb0c7.
We investigated the sensitivity of regional tumor response prediction to variability in voxel clustering techniques, imaging features, and machine learning algorithms in 25 patients with locally advanced non-small cell lung cancer (LA-NSCLC) enrolled on the FLARE-RT clinical trial. Metabolic tumor volumes (MTV) from pre-chemoradiation (PETpre) and mid-chemoradiation fluorodeoxyglucose-positron emission tomography (FDG PET) images (PETmid) were subdivided into K-means or hierarchical voxel clusters by standardized uptake values (SUV) and 3D-positions. MTV cluster separability was evaluated by CH index, and morphologic changes were captured by Dice similarity and centroid Euclidean distance. PETpre conventional features included SUVmean, MTV/MTV cluster size, and mean radiation dose. PETpre radiomics consisted of 41 intensity histogram and 3D texture features (PET Oncology Radiomics Test Suite) extracted from MTV or MTV clusters. Machine learning models (multiple linear regression, support vector regression, logistic regression, support vector machines) of conventional features or radiomic features were constructed to predict PETmid response. Leave-one-out-cross-validated root-mean-squared-error (RMSE) for continuous response regression (ΔSUVmean) and area-under-receiver-operating-characteristic-curve (AUC) for binary response classification were calculated. K-means MTV 2-clusters (MTVhi, MTVlo) achieved maximum CH index separability (Friedman p < 0.001). Between PETpre and PETmid, MTV cluster pairs overlapped (Dice 0.70-0.87) and migrated 0.6-1.1 cm. PETmid ΔSUVmean response prediction was superior in MTV and MTVlo (RMSE = 0.17-0.21) compared to MTVhi (RMSE = 0.42-0.52, Friedman p < 0.001). PETmid ΔSUVmean response class prediction performance trended higher in MTVlo (AUC = 0.83-0.88) compared to MTVhi (AUC = 0.44-0.58, Friedman p = 0.052). Models were more sensitive to MTV/MTV cluster regions (Friedman p = 0.026) than feature sets/algorithms (Wilcoxon signed-rank p = 0.36). Top-ranked radiomic features included GLZSM-LZHGE (large-zone-high-SUV), GTSDM-CP (cluster-prominence), GTSDM-CS (cluster-shade) and NGTDM-CNT (contrast). Top-ranked features were consistent between MTVhi and MTVlo cluster pairs but varied between MTVhi-MTVlo clusters, reflecting distinct regional radiomic phenotypes. Variability in tumor voxel cluster response prediction can inform robust radiomic target definition for risk-adaptive chemoradiation in patients with LA-NSCLC. FLARE-RT trial: NCT02773238.
我们研究了在 25 名局部晚期非小细胞肺癌（LA-NSCLC）患者中，肿瘤区域反应预测对体素聚类技术、成像特征和机器学习算法的变异性的敏感性，这些患者都参与了 FLARE-RT 临床试验。在进行放化疗前（PETpre）和放化疗中期（PETmid）氟代脱氧葡萄糖正电子发射断层扫描（FDG PET）图像中，代谢肿瘤体积（MTV）根据标准摄取值（SUV）和 3D 位置被细分为 K-均值或层次体素聚类。通过 CH 指数评估 MTV 聚类的可分离性，并通过 Dice 相似性和质心欧几里得距离捕获形态变化。PETpre 常规特征包括 SUVmean、MTV/MTV 聚类大小和平均辐射剂量。PETpre 放射组学由从 MTV 或 MTV 聚类中提取的 41 个强度直方图和 3D 纹理特征（PET Oncology Radiomics Test Suite）组成。对常规特征或放射组学特征构建了机器学习模型（多元线性回归、支持向量回归、逻辑回归、支持向量机），以预测 PETmid 反应。计算了连续反应回归（ΔSUVmean）的留一交叉验证均方根误差（RMSE）和二分类反应的受试者工作特征曲线下面积（AUC）。K-均值 MTV 2-聚类（MTVhi，MTVlo）达到了最大的 CH 指数可分离性（Friedman p<0.001）。在 PETpre 和 PETmid 之间，MTV 聚类对重叠（Dice 0.70-0.87），迁移了 0.6-1.1cm。与 MTVhi 相比（RMSE=0.42-0.52，Friedman p<0.001），MTV 和 MTVlo 中 PETmid ΔSUVmean 反应预测更优（RMSE=0.17-0.21）。与 MTVhi 相比（AUC=0.44-0.58，Friedman p=0.052），MTVlo 中 PETmid ΔSUVmean 反应分类预测性能呈上升趋势（AUC=0.83-0.88）。模型对 MTV/MTV 聚类区域（Friedman p=0.026）比特征集/算法（Wilcoxon 符号秩检验 p=0.36）更敏感。排名最高的放射组学特征包括 GLZSM-LZHGE（大区域高 SUV）、GTSDM-CP（聚类突出度）、GTSDM-CS（聚类阴影）和 NGTDM-CNT（对比度）。这些特征在 MTVhi 和 MTVlo 聚类对之间是一致的，但在 MTVhi-MTVlo 聚类之间是不同的，反映了不同的区域放射组学表型。肿瘤体素聚类反应预测的变异性可以为 LA-NSCLC 患者的风险适应性放化疗提供稳健的放射组学靶区定义。FLARE-RT 试验：NCT02773238。
