Abstract: Objective To construct a predictive model for the risk of postoperative cerebral edema in patients with meningioma based on preoperative radiomics and to evaluate its predictive value for the occurrence of postoperative cerebral edema.Methods A prospective study was conducted on 500 patients with meningioma who underwent surgical treatment in our hospital from May 2021 to March 2025.The patients were randomly divided into a training set (350 cases) and a validation set (150 cases) at a ratio of 7∶3.All patients underwent preoperative MRI examination.According to the presence or absence of postoperative cerebral edema, patients were divided into an occurrence group and a non-occurrence group.General data, laboratory examination results, and preoperative MRI radiomics features were analyzed between the two groups.Multivariate Logistic regression was used to analyze the influencing factors for postoperative cerebral edema in meningioma patients.A nomogram prediction model for postoperative cerebral edema was constructed and validated based on the influencing factors.The receiver operating characteristic (ROC) curve was used to evaluate the performance and clinical value of the prediction model, and internal validation of the nomogram was performed.Results In the training and validation sets, the occurrence group showed a higher proportion of irregular tumor morphology, absence of peritumoral brain edema, presence of dural tail sign, unclear tumor-brain interface, tumor located in the parasagittal/parafalcine region, tumor lesion>5 cm, and presence of feeding arteries compared to the non-occurrence group (P<0.05).The feature values of original_glszm_Gray Level Non Uniformity, wavelet-HLL_glszm_Gray Level Non Uniformity, exponential_glszm_Gray Level Non Uniformity, and log-sigma-3D_glszm_zone Entropy were significantly higher in the occurrence group than in the non-occurrence group(P<0.05).Multivariate Logistic regression analysis showed that regular tumormorphology, non-parasagittal/parafalcine tumor location, and tumor size (<5 cm) were independent protective factors against postoperative brain edema, while unclear tumor-brain interface, original_glszm_Gray Level Non Uniformity, wavelet-HLL_glszm_Gray Level Non Uniformity, exponential_glszm_Gray Level Non Uniformity, and log-sigma-3D_glszm_zone Entropy were independent risk factors for postoperative brain edema(P<0.05).The concordance index (C-index) of the nomogram model for predicting brain edema in the training set was 0.802.The calibration curve showed that the predicted probability of brain edema by the nomogram model was in good agreement with the actual incidence in the training set.Decision curve analysis (DCA) indicated that the prediction model had high net benefit when the threshold probability ranged from 20% to 60%, demonstrating certain clinical applicability.The receiver operating characteristic (ROC) curve showed that the area under the curve (AUC) of radiomic features for predicting postoperative brain edema in the validation set was 0.785 (95%CI: 0.671-0.874) (P<0.05).Conclusion The nomogram prediction model based on preoperative MRI radiomics features can effectively predict the risk of postoperative cerebral edema in patients with meningioma, providing strong support for clinical decision-making in patients with meningioma.

Key words: radiomics, meningioma, cerebral edema, predictive value