The method relies on deep learning models trained with ultra-high-resolution cerebellar MRI data (0.125 mm³) to capture the complex anatomy of cerebellar lobules. The framework combines neural networks with anatomical priors to automatically identify and segment cerebellar structures. The pipeline also integrates automated preprocessing and super-resolution reconstruction, allowing detailed cerebellar analysis even from standard MRI scans. The output provides high-resolution cerebellar segmentation and volumetric measurements of cerebellar lobules.