Consistency and grade prediction of intracranial meningiomas based on fractal geometry analysis

Meningiomas are the most common primary tumors in the central nervous system. Surgical resection remains the main treatment option, often resulting in a curative outcome; however, careful preoperative planning is essential. One of the primary concerns for neurosurgeons treating meningiomas is tumor consistency, as this has a significantly impact on the likelihood of complete resection. Predicting the consistency and histology of a meningioma prior to surgery is valuable for selecting the appropriate surgical instruments and planning the approach. We conducted a retrospective study to analyze clinical data and preoperative MRI images of patients who underwent surgery for intracranial meningiomas. T1, T1c, T2, and FLAIR sequences were obtained for all patients. Surgical notes were reviewed to assess tumor consistency. Tumor segmentation was performed using ITK-SNAP software. Fractal analysis and statistical analyses were made, including t-tests, Fisher's exact tests, logistic regression, and ROC analysis. Forty-eight patients met the selection criteria. For prediction of consistency when only fractal parameters were used, lacunarity index was able to discriminate between soft and hard consistency with an AUC value of 0.745 (95% CI: 0.538-0.958). When tumor homogeneity was added, these values changed to 0.763 (95% CI: 0.518-1.000). For prediction of histological grade, an AUC value of 0.697 (95% CI: 0.490-0.952) was found, using only fractal dimension. When age, tumor homogeneity and volume parameters were added, this value increased to 0.841 (95% CI: 0.625-1.000). Our study suggests that fractal metrics are useful tools for preoperative estimation of tumor consistency and histological grading.