MRI-Based Radiomics to Differentiate between Benign and Malignant Parotid Tumors With External Validation

• DOI: 10.3389/fonc.2021.656918
• Publication Date: 2021-04-27T09:01:22Z
• Authors: Francesca Piludu; Simona Marzi; Marco Ravanelli; Raul Pellini; Renato Covello; Irene Terrenato; Davide Farina; Riccardo Campora; Valentina Ferrazzoli; Antonello Vidiri
• Keywords: Adenolymphoma

The differentiation between benign and malignant parotid lesions is crucial to defining the treatment plan, which highly depends on tumor histology. We aimed evaluate role of MRI-based radiomics using both T2-weighted (T2-w) images Apparent Diffusion Coefficient (ADC) maps in lesions, order develop predictive models with an external validation cohort.A sample 69 untreated was evaluated retrospectively, including 37 (of 13 were Warthin's tumors) 32 tumors. patient population divided into three groups: (24 cases), (13 (32 compared pairs. First- second-order features derived for each lesion. Margins contrast enhancement patterns (CE) qualitatively assessed. model final feature set achieved support vector machine binary classification algorithm.Models discriminating tumors, tumors had accuracy 86.7%, 91.9% 80.4%, respectively. After selection process, four parameters used, histogram-based from ADC T2-w images, shape-based types margins and/or CE. Comparable accuracies obtained after cohort.Radiomic analysis ADC, qualitative scores evaluating CE allowed us obtain good excellent diagnostic differentiating confirmed cohort.