Multimodal MRI radiomics for prediction of 1p/19q co-deletion status in low grade gliomas

Motivation: A non-invasive method for predicting the status of glioma 1p/19q is indeed, as some patients, due to advanced age, poor neurological condition, or tumor location, are not suitable candidates for surgery or biopsy. Goal(s): To develop radiomics-based predictive models that can accurately determine the 1p/19q codeletion status in lower-grade gliomas. Approach: Four radiomics models were developed and analyzed: LR, LRlasso, SVM, and LDA, and the predictive performance of these models was validated. Results: The four radiomics models predicted 1p/19q missing code status with high accuracy and diagnostic performance. This suggests that these models hold promise as reliable non-invasive tools in glioma diagnosis. Impact: This study underscores a radiomics model based on multimodal MRI (T2WI, T1WI, FLAIR, CE-T1WI, and DWI) to non-invasively predict the 1p/19q codeletion status in diffuse low-grade gliomas pre-surgery. The model provides reliable evidence for personalized treatment of patients.