Machine learning-based radiomics analysis in predicting the meningioma grade using multiparametric MRI

Area of Science:

• Radiology
• Medical Imaging
• Oncology

Background:

• Meningiomas are the most common primary intracranial tumors.
• Accurate grading of meningiomas is crucial for treatment planning and prognosis.
• Multiparametric MRI offers detailed tissue characterization for tumor assessment.

Purpose of the Study:

• To evaluate the predictive performance of radiomic models derived from multiparametric MRI for meningioma grading.
• To compare the efficacy of different combinations of MRI sequences in predicting tumor grade.

Main Methods:

• Radiomic features were extracted from conventional MRI (cMRI), ADC maps, and SWI in 316 patients.
• Various radiomic models were constructed using feature selection (LASSO) and classification (Random Forest).
• Model performance was assessed using receiver operating characteristic (ROC) curves and area under the curve (AUC) with nested leave-one-out cross-validation (LOOCV).

Main Results:

• The combined cMRI + ADC + SWI radiomic model achieved the highest AUC of 0.84 without subsampling and 0.81 with subsampling.
• Other models showed AUCs ranging from 0.71 to 0.80.
• The combined model significantly outperformed individual sequences in predicting meningioma grade.

Conclusions:

• A multiparametric radiomic model integrating cMRI, ADC map, and SWI demonstrates superior performance in predicting meningioma grade.
• This approach holds promise for improving clinical decision-making in meningioma management.
• Radiomics can enhance the non-invasive assessment of meningioma aggressiveness.