Automatic segmentation and classification of GBM and SBM using a 3D deep learning model on multiparametric MRI: a multi-center study

Motivation: Differentiating between glioblastoma (GBM) and solitary brain metastasis (SBM) before surgery is challenging yet essential for clinical decisions. Goal(s): To construct a three-dimensional (3D) deep learning (DL) model for the automated segmentation and classification of GBM and SBM based on multiparametric MRI. Approach: 314 patients from multiple medical centers were recruited. Tumor segmentation was performed using no-new-UNet, followed by the development and comparison of 3D DL, 2D DL, and radiomics models for classification. Additionally, three radiologists with varying experience levels conducted a two-round classification analysis, with and without a 3D DL model as a reference. Results: 3D DL model showed the best performance. Impact: Reading with 3D DL model improves the diagnostic accuracy of radiologists, thereby enhancing patient treatment and prognosis potentially. Considering its promising performance, it is recommended for routine clinical application.