NCCT-based habitats and peri-tumoral radiomics enhance pre-operative prediction of benign and malignant sacral tumors

Abstract Background Differentiating between benign and malignant sacral tumors is essential for guiding treatment strategies. This study aims to develop a novel approach utilizing non-contrast computed tomography (NCCT) for the accurate classification of sacral tumors. Methods From July 2021 to September 2023, we retrospectively included 112 patients with pathologically confirmed sacral tumors. Tumor regions of interest (ROIs) were segmented into three spatial habitats using K-means clustering and dilated to thicknesses of 3 mm and 6 mm. A comprehensive analysis of the peri-tumoral and habitat-defined regions was conducted, utilizing features extracted from NCCT images. An ensemble methodology integrating clinical indicators with the habitats and peri-tumoral models was employed to develop a fusion model, termed the ensemble nomogram. The efficacy of these models was assessed using the area under the receiver operating characteristics curve (AUC), calibration curves, and decision curve analysis. Results The Multi-Layer Perceptron (MLP), based on ROI 3mm , exhibited enhanced performance relative to that utilizing ROI 6mm . The MLP-based habitats model achieved an AUC of 0.877 in the test cohorts, outperforming the ROI intra -based model. The ensemble nomogram, integrating age, peri-tumoral, and habitats models, attained the highest performance with an AUC of 0.892 in the test sets, significantly outperforming the clinical model (AUC = 0.619, P = 0.02). The calibration curve of the nomogram demonstrated good concordance between predictions and observed outcomes, and DCA revealed a high overall net benefit. Conclusion NCCT-based habitats and peri-tumoral radiomics improve preoperative prediction of sacral tumors, with habitat signatures offering precise predictions to guide personalized treatment.