Differentiation Between Benign and Nonbenign Meningiomas by Using Texture Analysis From Multiparametric MRI

Abstract

BackgroundIt is difficult to prospectively differentiate between benign (World Health Organization [WHO] I) and nonbenign (WHO II and III) meningiomas.PurposeTo evaluate the feasibility of preoperative differentiation between benign and nonbenign meningiomas by using texture analysis from multiparametric MR data.Study TypeRetrospective.SubjectsIn all, 184 patients with meningioma (139 benign and 45 nonbenign) were included as the training cohort and 79 patients with meningioma (60 benign and 19 nonbenign) were included as the external validation cohort.Field Strength/SequenceT1‐weighted, T2‐weighted, and contrast‐enhanced T1‐weighted imaging were performed on 1.5 or 3.0T MR systems from two centers.AssessmentTumor segmentation and radiological characteristic (RC) evaluation were performed by experienced radiologists. The texture features were extracted from preprocessed images and combined with RCs, and then the combined features were reduced by using a two‐step feature selection. Three single‐sequence models and a multiparametric MRI (the combination of single sequences) model were constructed and then evaluated with the external validation cohort.Statistical TestsArea under receiver operating characteristic curve (AUC), accuracy (Acc), f1‐score (F1), sensitivity (Sen), and specificity (Spec), were calculated to quantify the performance of the models.ResultsAmong the four texture models, the multiparametric MRI model demonstrated the best performance for differentiating between benign and nonbenign meningiomas in both the training and external validation cohorts (AUC 0.91, Acc 89%, F1 0.88, Sen 0.93, and Spec 0.87 in the training cohort; AUC 0.83, Acc 80%, F1 0.77, Sen 0.84, and Spec 0.78 in the validation cohort).Data ConclusionNonbenign meningiomas might be preoperatively differentiated from benign meningiomas by using texture analysis from multiparametric MR data.Level of Evidence: 3Technical Efficacy Stage: 2J. Magn. Reson. Imaging 2020;51:1810–1820.

Meningioma MRI radiomics and machine learning: systematic review, quality score assessment, and meta-analysis

“…This finding supports the use of multiple imaging sequences rather than relying exclusively on T1 contrast-enhanced sequences for future investigations. Similarly, the good accuracy (AUC = 0.89) obtained by studies (n = 6) who included image pre-processing in their pipeline also suggests the usefulness of this step [11,33,37,45,48,49]. While the AUC for single institution (n = 4) and multicenter studies was equally high (AUC = 0.88), external testing of ML models is always preferable to demonstrate their ability to generalize.…”

“…Despite this, its AUC value is among the lowest (0.78) suggesting that these may not be essential in the preoperative definition of meningioma grading. It is also interesting to note that most (n = 5) of the studies used linear ML models [11,31,37,48,49] while only one included a data augmentation technique [33].…”

“…Similarly, while k-fold cross-validation helps in extracting more information and reliable results from small datasets, its exclusive use may present some issues as there is no final model whose performance can be tested on unseen data. In all, 4 studies only employed cross-validation, with better results than the remaining (AUC = 0.92 vs 0.84) [11,40,48,49]. Ideally, it would be preferable to use cross-validation for model tuning and initial testing followed by further assessment on new data, as done in 2 cases (AUC = 0.82 and 0.83).…”

“…Ideally, it would be preferable to use cross-validation for model tuning and initial testing followed by further assessment on new data, as done in 2 cases (AUC = 0.82 and 0.83). This approach combines the advantages of both testing strategies [48,49].…”

“…One study scored an unclear risk of bias in index test domain as the radiomics feature extraction was performed from both diffusion-weighted images (DWI) and apparent diffusion coefficient (ADC) maps [37]. Finally, the time elapsed between MRI and neurosurgery was not reported in 6 studies, thus scoring an unclear risk of bias in the flow and timing domain [31][32][33][34][35][36][37][38][39][40][41][42][43][44][45]49]. All the studies included in the meta-analysis had low concerns regarding applicability for the three domains (patient selection, index test, and reference standard).…”