Identification of High-Risk Atypical Meningiomas According to Semantic and Radiomic Features

Abstract

Up to 60% of atypical meningiomas (World Health Organization (WHO) grade II) reoccur within 5 years after resection. However, no clear radiological criteria exist to identify tumors with higher risk of relapse. In this study, we aimed to assess the association of certain radiomic and semantic features of atypical meningiomas in MRI with tumor recurrence. We identified patients operated on primary atypical meningiomas in our department from 2007 to 2017. An analysis of 13 quantitatively defined radiomic and 11 qualitatively defined semantic criteria was performed based on preoperative MRI scans. Imaging characteristics were assessed along with clinical and survival data. The analysis included 76 patients (59% women, mean age 59 years). Complete tumor resection was achieved in 65 (86%) cases, and tumor relapse occurred in 17 (22%) cases. Mean follow-up time was 41.6 (range 3–168) months. Cystic component was significantly associated with tumor recurrence (odds ratio (OR) 21.7, 95% confidence interval (CI) 3.8–124.5) and shorter progression-free survival (33.2 vs. 80.7 months, p < 0.001), whereas radiomic characteristics had no predictive value in univariate analysis. However, multivariate analysis demonstrated significant predictive value of high cluster prominence (hazard ratio (HR) 5.89 (1.03–33.73) and cystic component (HR 20.21 (2.46–166.02)) for tumor recurrence. The combination of radiomic and semantic features might be an effective tool for identifying patients with high-risk atypical meningiomas. The presence of a cystic component in these tumors is associated with a high risk of tumor recurrence.

Pre-operative MRI Radiomics for the Prediction of Progression and Recurrence in Meningiomas

“…It works by extracting a large number of quantitative characteristics from a medical image and then analyses these features by means of a series of machine learning algorithms (11). Although the radiomics approach for the evaluation of meningiomas pertaining to tumor grades and histological subtypes had recently been reported (12)(13)(14)(15), models for predicting clinical outcomes in overall meningiomas are still rare (10,16). Among the machine learning techniques, several studies had reported that support vector machine (SVM) classifiers offer excellent results in the classification and segmentation in brain tumors (17)(18)(19)(20)(21)(22).…”

Exploring MRI Characteristics of Brain Diffuse Midline Gliomas With the H3 K27M Mutation Using Radiomics

“…This was not the first time for cyst formation to be scrutinized. In a previous study, cyst formation was the only feature selected among 13 radiomic and 11 visually accessible features for the identification of high-risk atypical meningiomas ( 34 ). Another previous study ( 8 ) showed that although there was no significant difference in the cystic component or necrosis between DMG-M and DMG-W tumors, there was a higher ratio of cystic components or necrosis of DMG-M (62.5%) compared to DMG-M (33.3%).…”

Multi-Slice Radiomic Analysis of Apparent Diffusion Coefficient Metrics Improves Evaluation of Brain Alterations in Neonates With Congenital Heart Diseases

“…This may be due to substantial overlap in the regional ADCs for whole slices or whole brains that might lead to a lack of significant differences in texture features. Recent advances combined ADC metrics and texture features to perform classification for high-risk atypical meningiomas (35). Compared to ADC metrics that showed significance, texture features was suggested to have limited contribution to tumor classification.…”