Multicenter study demonstrates radiomic features derived from magnetic resonance perfusion images identify pseudoprogression in glioblastoma

Abstract

Pseudoprogression (PsP) is a diagnostic clinical dilemma in cancer. In this study, we retrospectively analyse glioblastoma patients, and using their dynamic susceptibility contrast and dynamic contrast-enhanced perfusion MRI images we build a classifier using radiomic features obtained from both Ktrans and rCBV maps coupled with support vector machines. We achieve an accuracy of 90.82% (area under the curve (AUC) = 89.10%, sensitivity = 91.36%, 67 specificity = 88.24%, p = 0.017) in differentiating between pseudoprogression (PsP) and progressive disease (PD). The diagnostic performances of the models built using radiomic features from Ktrans and rCBV separately were equally high (Ktrans: AUC = 94%, 69 p = 0.012; rCBV: AUC = 89.8%, p = 0.004). Thus, this MR perfusion-based radiomic model demonstrates high accuracy, sensitivity and specificity in discriminating PsP from PD, thus provides a reliable alternative for noninvasive identification of PsP versus PD at the time of clinical/radiologic question. This study also illustrates the successful application of radiomic analysis as an advanced processing step on different MR perfusion maps.

Fig. 1

Image post-processing radiomic workflow. a relative cerebral blood volume (rCBV) and K_trans maps of perfusion MRI are acquired. b Segmentation of the region of interest using 3D slicer software. c Radiomic feature extraction from the whole tumor volume. d Statistical analysis: radiomic and clinical features are analyzed to determince their diagnostic and predictive values

Fig. 2

Model building and evaluation using the selected K_trans features (60 features). a, b ROC curve depicts the predictive model building using C5.0 (P-value 1.512e−11) and SVM methods (P-value 0.003744) respectively. c, d 10-fold cross-validation ROC curve (P-value 1.512e−11) and Leave-One-Out Cross-Validation (LOOCV) ROC curve (P-value 0.004) depicts the performance of the model

Fig. 3

Model building and evaluation using the selected rCBV features (160 features). a, b ROC curve depicts the predictive model building using C5.0 (P-value 1.512e−11) and SVM methods (P-value 0.012) respectively. c, d 10-fold cross-validation ROC curve (P-value 1.512e−11) and Leave-One-Out Cross-Validation (LOOCV) ROC curve (P-value 0.012) depicts the performance of the model

Fig. 4

Model building and evaluation using the selected merged K_trans and rCBV features (60 features). a, b ROC curve depicts the predictive model building using C5.0 (P-value 1.512e−11) and SVM methods (P-value 0.017), respectively. c, d 10-fold cross-validation ROC curve (P-value 1.512e−11) and Leave-One-Out Cross-Validation (LOOCV) ROC curve (P-value 0.02) depicts the performance of the model