Quick Start

Prerequisites

• Python 3.8+
• MONAI Label
• 3D Slicer or OHIF Viewer

Installation

# Install MONAI Label
pip install monailabel-weekly

# Clone the model repository
git clone https://github.com/Hirriririir/Multimodal-Multiethnic-Thigh-Muscle-MRI-analysis.git
cd Multimodal-Multiethnic-Thigh-Muscle-MRI-analysis

Start MONAI Label Server

monailabel start_server \
    --app . \
    --studies /path/to/your/mri/data \
    --conf models segresnet

Segmentation Workflow

1. Open 3D Slicer and connect to the MONAI Label server
2. Load a thigh MRI scan from the server
3. Click Auto Segmentation to run the model
4. Review and refine the 11-muscle segmentation result
5. Export segmentation masks for downstream analysis

Data Download

Release Data

The following files are available on the GitHub Release v1.0:

File	Size	Description
pretrained_segmentation_muscle.pt	~329 MB	Pretrained SegResNet model weights
HuashanMyo.zip	~746 MB	Huashan Hospital multimodal thigh MRI dataset
Folkhalsan.zip	~150 MB	Folkhälsan Research Center thigh MRI dataset
Dataset.json	~42 KB	MONAI Label dataset configuration

ITK-SNAP Label File

Download the ITK-SNAP label file for 11-muscle annotation:

ThighMuscle.label

Load in ITK-SNAP via Segmentation → Import Label Descriptions.

Analysis

The repository includes Jupyter notebooks for:

• Radiomics feature extraction and correlation analysis
• Fat fraction quantification from IDEAL sequences
• Multi-ethnic metric comparison across cohorts
• Statistical analysis scripts (R-based correlation matrices)