SegMamba-V2
A general 3D medical image segmentation backbone for long-range volumetric modeling across organs and modalities.
Overview
SegMamba-V2 is a general-purpose 3D segmentation backbone built to capture long-range dependencies in volumetric medical images without the computational cost profile of standard transformer pipelines. The model extends Mamba-based sequential modeling into 3D space with tri-orientated spatial blocks and hierarchical downsampling.
In the lab’s imaging and surgical direction, this project functions as a core perception backbone: a reusable model for anatomical understanding across colorectal cancer and other volumetric segmentation settings.