Fully automated analysis combining [18F]-FET-PET and multiparametric MRI including DSC perfusion and APTw imaging: a promising tool for objective evaluation of glioma progression

Abstract

Purpose: To evaluate diagnostic accuracy of fully automated analysis of multimodal imaging data using [¹⁸F]-FET-PET and MRI (including amide proton transfer-weighted (APTw) imaging and dynamic-susceptibility-contrast (DSC) perfusion) in differentiation of tumor progression from treatment-related changes in patients with glioma.

Material and methods: At suspected tumor progression, MRI and [¹⁸F]-FET-PET data as part of a retrospective analysis of an observational cohort of 66 patients/74 scans (51 glioblastoma and 23 lower-grade-glioma, 8 patients included at two different time points) were automatically segmented into necrosis, FLAIR-hyperintense, and contrast-enhancing areas using an ensemble of deep learning algorithms. In parallel, previous MR exam was processed in a similar way to subtract preexisting tumor areas and focus on progressive tumor only. Within these progressive areas, intensity statistics were automatically extracted from [¹⁸F]-FET-PET, APTw, and DSC-derived cerebral-blood-volume (CBV) maps and used to train a Random Forest classifier with threefold cross-validation. To evaluate contribution of the imaging modalities to the classifier's performance, impurity-based importance measures were collected. Classifier performance was compared with radiology reports and interdisciplinary tumor board assessments.

Results: In 57/74 cases (77%), tumor progression was confirmed histopathologically (39 cases) or via follow-up imaging (18 cases), while remaining 17 cases were diagnosed as treatment-related changes. The classification accuracy of the Random Forest classifier was 0.86, 95% CI 0.77-0.93 (sensitivity 0.91, 95% CI 0.81-0.97; specificity 0.71, 95% CI 0.44-0.9), significantly above the no-information rate of 0.77 (p = 0.03), and higher compared to an accuracy of 0.82 for MRI (95% CI 0.72-0.9), 0.81 for [¹⁸F]-FET-PET (95% CI 0.7-0.89), and 0.81 for expert consensus (95% CI 0.7-0.89), although these differences were not statistically significant (p > 0.1 for all comparisons, McNemar test). [¹⁸F]-FET-PET hot-spot volume was single-most important variable, with relevant contribution from all imaging modalities.

Conclusion: Automated, joint image analysis of [¹⁸F]-FET-PET and advanced MR imaging techniques APTw and DSC perfusion is a promising tool for objective response assessment in gliomas.

Keywords: APTw; B. coshared last; DSC perfusion; Fully automated; Glioma progression; J. S. and Wiestler; Kirschke; Multiparametric MRI; [18F]-FET-PET.

Fig. 1

Flow chart patient selection: From initial 163 [¹⁸F]-FET-PET data sets performed to differentiate PD vs. TRC, we excluded 89 data sets due to missing data or insufficient image quality

Fig. 2

ROC curve of the Random Forest (RF) classifier, derived from the fully automated multimodal evaluation (AUC = 0.85) and a comparison of the point of operation with MRI, PET, and the expert consensus

Fig. 3

Feature importance. The length of the bar indicates relative importance (summing to 1) of each input feature for classifier performance, i.e., longer bars indicate relatively higher importance

Fig. 4

Example images of progressive disease in a 65-year-old female patient with left frontal GBM and a new contrast enhancement superior to the former resection area (upper left: ce T1-w, lower left: previous time point, upper middle: automated segmentation overlay (green: new FLAIR edema area; yellow: new ce area), lower middle: CBV, upper right: [¹⁸F]-FET-PET, lower right APTw)

Fig. 5

Example images of TRC in a 34-year-old male with a new contrast enhancing focus next to the resection cavity after therapy of a left frontal GBM (upper left: ce T1-w, lower left: previous time point, upper middle: automated segmentation overlay (green: new FLAIR edema area; yellow: new ce area), lower middle: CBV, upper right: [¹⁸F]-FET-PET, lower right APTw)