Synthesizing healthy brain scans from diseased brain scans offers a potential solution to address the limitations of general-purpose algorithms, such as tissue segmentation and brain extraction algorithms, which may not effectively handle diseased images. We consider this a 3D inpainting task and investigate the adaptation of 2D inpainting methods to meet the requirements of 3D magnetic resonance imaging(MRI) data. Our contributions encompass potential modifications tailored to MRI-specific needs, and we conducted evaluations of multiple inpainting techniques using the BraTS2023 Inpainting datasets to assess their efficacy and limitations.