Multimodal MRI Brain Tumor Segmentation Using 3D and 3D/2D Methods: A Study on the MICCAI BRATS Dataset

This research paper presents a comparative analysis of the performance of 3D and 3D/2D brain tumor segmentation methods using DPSO on the BRATS 2019 dataset. The study specifically focuses on two categories: High-Grade Glioma (HGG) and Low-Grade Glioma (LGG). In the HGG category, the 3D segmentation method achieved moderate overlap with dice coefficients of 72% and 80% in Datasets 1 and 2, respectively. Comparatively, the 3D/2D method demonstrated slightly lower performance with dice coefficients of 70% and 77%. Shifting to the LGG category, the 3D method exhibited high agreement with dice coefficients of 90.20% and 95% in Datasets 3 and 4, respectively. The 3D/2D method achieved slightly lower dice coefficients of 87% and 84% in the same datasets. Overall, the results emphasize the superiority of the 3D method in accurately delineating brain tumors, as it consistently outperformed the 3D/2D method across all datasets. However, it is important to note that the 3D/2D method also demonstrated reasonably good performance, making it a viable alternative, particularly in scenarios where computational resources or time constraints are a consideration. This research significantly contributes to the advancement of brain tumor segmentation methods by offering valuable insights into the comparative performance of 3D and 3D/2D approaches. The findings shed light on the strengths and limitations of each method.