UMambaAdj: Advancing GTV Segmentation for Head and Neck Cancer in MRI-Guided RT with UMamba and nnU-Net ResEnc Planner

This repository contains the trained weights and validation results of the proposed methods for T2-weighted MRI head and neck tumor segmentation, including GTVp and GTVn segmentation for the HNTS-MRG 2024 challenge.

Preprocessing, postprocessing and model codes can be found at UMambaAdj Github.

Available Model Weights

The trained weights and validation results are stored in the following directories:

• nnUNetTrainerResenc__nnUNetResEncUNetMPlans__3d_fullres_bs4
• nnUNetTrainerUmamba__nnUNetResEncUNetMPlans__3d_fullres_bs4

These directories correspond to: \1. nnUNetTrainerResenc: The nnU-Net Residual Encoder model with M plans. \2. nnUNetTrainerUmamba: The UMamba model with the proposed modifications.

How to Use

Download the trained weights from this repository. Load the model weights into your nnU-Net environment following the standard loading instructions provided by nnU-Net.

For more details on the validation performance, refer to the HNTS-MRG 2024 challenge and the paper.