Feasibility of radiomic feature harmonization for pooling of [18F]FET or [18F]GE-180 PET images of gliomas.

INTRODUCTION: Large datasets are required to ensure reliable non-invasive glioma assessment with radiomics-based machine learning methods. This can often only be achieved by pooling images from different centers. Moreover, trained models should perform with high accuracy when applied to data from different centers. In this study, the impact of reconstruction settings and segmentation methods on radiomic features derived from amino acid and TSPO PET images of glioma patients was examined. Additionally, the ability to model and thus reduce feature differences was investigated. METHODS: [18F]FET and [18F]GE-180 PET data were acquired from 19 glioma patients. For each acquisition, 10 reconstruction settings and 9 segmentation methods were included to emulate multicentric data. Statistical robustness measures were calculated before and after ComBat harmonization. Differences between features due to setting variations were assessed using Friedman test, coefficient of variation (CV) and inter-rater reliability measures, including intraclass and Spearman's rank correlation coefficients and Fleiss' Kappa. RESULTS: According to Friedman analyses, most features (>60%) showed significant differences. Yet, CV and inter-rater reliability measures indicated higher robustness. ComBat resulted in almost complete harmonization (>87%) according to Friedman test and little to no improvement according to CV and inter-rater reliability measures. [18F]GE-180 features were more sensitive to reconstruction settings than [18F]FET features. CONCLUSIONS: According to Friedman test, feature distributions could be successfully aligned using ComBat. However, depending on settings, changes in patient ranks were observed for some features and could not be eliminated by harmonization. Thus, for clinical utilization it is recommended to exclude affected features.

Improved Detection of Transosseous Meningiomas Using 68Ga-DOTATATE PET/CT Compared with Contrast-Enhanced MRI.

68Ga-DOTATATE PET/CT enables detection of meningioma tissue based on somatostatin receptor 2 expression. Transosseous extension of intracranial meningiomas is known to be an important risk factor for tumor recurrence and patient mortality. We analyzed the diagnostic performance of 68Ga-DOTATATE PET/CT and contrast-enhanced MRI (CE-MRI) for the detection of osseous infiltration using qualitative and quantitative imaging parameters. Methods: In this institutional review board-approved retrospective study, subjects were selected from 327 consecutive 68Ga-DOTATATE PET/CT examinations for evaluation of confirmed or suspected meningioma. Inclusion criteria were CE-MRI within 30 d and pathology-confirmed meningioma diagnosis with inclusion or exclusion of transosseous extension as the standard of reference. Imaging was analyzed by two readers. Tracer uptake values and meningioma volumes were determined. χ2, Mann-Whitney U, Wilcoxon signed rank, and McNemar tests, as well as receiver-operating-characteristic analyses, were performed to compare variables and diagnostic performance. Results: Eighty-two patients fulfilled the inclusion criteria. Patients with transosseous extension of meningioma (n = 67) showed significantly larger lesions (median, 12.8 vs. 3.3 mL; P < 0.001) and significantly higher tracer uptake values (median SUVmax, 14.2 vs. 7.6; P = 0.011) than patients with extraosseous meningiomas (n = 15). 68Ga-DOTATATE PET/CT in comparison to CE-MRI performed at a higher sensitivity (98.5% vs. 53.7%) while maintaining high specificity (86.7% vs. 93.3%) in the detection of osseous involvement (P < 0.001). In receiver-operating-characteristic analysis, PET/CT assessment performed better than CE-MRI (area under the curve, 0.932 vs. 0.773). PET/CT- and CE-MRI-based volume estimation yielded comparable results for extraosseous meningiomas (P = 0.132) and the extraosseous part of transosseous meningiomas (P = 0.636), whereas the volume of the intraosseous part was assessed as significantly larger by PET/CT (P < 0.001). Conclusion:68Ga-DOTATATE PET/CT enables improved detection of the transosseous extension of intracranial meningiomas compared with CE-MRI.