Meningioma Consistency Can Be Defined by Combining the Radiomic Features of Magnetic Resonance Imaging and Ultrasound...

Area of Science:

• Neurosurgery
• Radiology
• Medical Imaging Analysis

Background:

• Meningioma consistency impacts surgical planning and resection extent.
• Accurate prediction of tumor consistency is crucial for optimizing surgical approaches.

Purpose of the Study:

• To develop a predictive model for meningioma consistency.
• Utilize radiomic features from preoperative MRI and intraoperative ultrasound elastography (IOUS-E).

Main Methods:

• Retrospective analysis of 18 supratentorial meningiomas.
• Radiomic feature extraction from MRI (T1, ADC, T2).
• Machine learning models (Naive Bayes, Random Forest, SVM, etc.) with feature selection.

Main Results:

• A Naive Bayes model combined with Information Gain and ReliefF filters achieved 0.961 AUC.
• Classification accuracy reached 94% for predicting hard or soft meningiomas.
• Key radiomic features derived from T1-post-contrast, ADC map, and T2-weighted images.

Conclusions:

• A high-precision classification model for meningioma consistency was developed.
• The model leverages radiomic features from preoperative MRI.
• This tool can aid in preoperative surgical planning for meningiomas.