[18F]FDG-PET/CT radiomics for the identification of genetic clusters in pheochromocytomas and paragangliomas

• DOI: 10.1007/s00330-022-09034-5
• Publication Date: 2022-08-24T11:06:10Z
• Authors: Wyanne A. Noortman; Dennis Vriens; Lioe-Fee de Geus-Oei; Cornelis H. Slump; Erik H. Aarntzen; Anouk van Berkel; Henri J. L. M. Ti...; Floris H. P. van ...
• Keywords: Neuroradiology; Interventional radiology; Identification

Based on germline and somatic mutation profiles, pheochromocytomas paragangliomas (PPGLs) can be classified into different clusters. We investigated the use of [18F]FDG-PET/CT radiomics, SUVmax biochemical profile for identification genetic clusters PPGLs.In this single-centre cohort, 40 PPGLs (13 cluster 1, 18 2, 9 sporadic) were delineated using a 41% adaptive threshold SUVpeak ([18F]FDG-PET) manually (low-dose CT; ldCT). Using PyRadiomics, 211 radiomic features extracted. Stratified 5-fold cross-validation was performed multinomial logistic regression with dimensionality reduction incorporated per fold. Classification performances biochemistry, PET(/CT) models compared presented as mean (multiclass) test AUCs over five folds. Results validated sham experiment, randomly shuffling outcome labels.The model biochemistry only could identify (multiclass AUC 0.60). The three-factor PET had best classification performance 0.88). A simplified almost similarly. Addition ldCT decreased performances. All approximately 0.50.PET radiomics achieves better to SUVmax, combined approaches, especially differentiation sporadic PPGLs. Nevertheless, alone might preferred clinically, weighing against laborious analysis. limited added value overall PPGL should in larger external cohort.• Radiomics derived from has potential improve paragangliomas. • Cluster 1 2 generally present distinctive characteristics that captured [18F]FDG-PET imaging. Sporadic appear more heterogeneous, frequently resembling occasionally