MRI Radiomic Features to Predict IDH1 Mutation Status in Gliomas: A Machine Learning Approach using Gradient Tree...

Area of Science:

• Neuro-oncology
• Radiology
• Machine Learning
• Medical Imaging

Background:

• Isocitrate dehydrogenase 1 (IDH1) mutation status is a critical prognostic indicator in glioma patients.
• Machine learning (ML) shows potential in analyzing complex medical imaging data for disease characterization.
• Predicting IDH1 status non-invasively could improve patient management and treatment strategies.

Purpose of the Study:

• To investigate the efficacy of ML analysis of multiparametric radiomic features from preoperative MRI in predicting IDH1 mutation status in glioma.
• To evaluate the performance of ML models trained on different MRI sequences (FLAIR, DWI) and combined sequences.

Main Methods:

• Retrospective study of 100 glioma patients with known IDH1 status and preoperative MRI scans.
• Extraction of radiomic features from Fluid-Attenuated Inversion Recovery (FLAIR) and Diffused Weighted Imaging (DWI) sequences.
• Training and hyperparameter tuning of eXtreme Gradient Boosting (XGBoost) classifiers, with data augmentation using Synthetic Minority Oversampling Technique (SMOTE).

Main Results:

• The DWI-trained XGBoost model achieved an Area Under the Curve (AUC) of 0.97, 90% accuracy, and 0.75 f1-score on the test set.
• The FLAIR-trained XGBoost model demonstrated comparable performance with an AUC of 0.95, 90% accuracy, and 0.75 f1-score.
• Combining FLAIR and DWI features did not significantly improve the predictive accuracy of the ML models.

Conclusions:

• ML analysis of multiparametric radiomic features from preoperative MRI can predict IDH1 mutation status in glioma with high accuracy (>90%).
• DWI and FLAIR sequences, when analyzed independently using XGBoost, are effective predictors of IDH1 status.
• This radiomic approach offers a promising, non-invasive method for assessing IDH1 mutation status in glioma patients.