Machine learning based MRI radiomics model in predicting postoperative severe poor outcomes after resection of meningioma.

Motivation: Meningioma is the most common intracranial tumor, and resection is the standard treatment. Postoperative severe poor outcomes (PSPO) remain a major factor significantly affecting patients' quality of life. Goal(s): To develop a novel radiomics model that can accurately predict PSPO after meningioma resection. Approach: Reviewed 148 patients with pathology-confirmed meningiomas, extracting radiomics features from tumor enhancement and peritumoral edema regions to build machine learning-based predictive models. Results: The combined model, incorporating both tumor enhancement and peritumoral edema radiomics features, along with clinical characteristics demonstrated the best predictive performance with AUC values of 0.88 and 0.87 in the training and validation sets. Impact: The novel model can non-invasively predict PSPO after meningioma resection, enabling early identification of high-risk patients. This approach can optimize clinical decision-making and enhance postoperative management, ultimately improving outcomes for meningioma patients.