Auto-FWLL-SSM (LLSSM)

Automated Female Whole Lower Limb Statistical Shape Model

Official implementation of the paper:

Multi-structure statistical shape model of the lower limb musculoskeletal system in patients with hip osteoarthritis

Yuto Masaki, Yoshito Otake, Mazen Soufi, Keisuke Uemura, Masaki Takao, Takuma Miyamoto, Yasuhito Tanaka, Seiji Okada, Nobuhiko Sugano, Yoshinobu Sato

Under review

Overview

Auto-FWLL-SSM is a Statistical Shape Model (SSM) of the entire lower limb musculoskeletal system, constructed from CT scans of 560 Japanese female patients with hip osteoarthritis (HOA). The model integrates 60 anatomical structures including bones, muscles, and skin using deep learning-based automated segmentation and PCA-based shape modeling.

This repository provides tools for:

• Reading and writing multi-structure polygon meshes (VTK format)
• Decomposing labeled meshes into individual anatomical structures
• Applying SSM-based shape deformation using PCA components

Requirements

• Python 3.x
• NumPy
• VTK

Optional

• PyTorch (for GPU/batched SSM operations)
• SimpleITK (for reading .mha/.mhd/.nii.gz files)

SSM Models

Under construction

Usage

# Read a labeled mesh and export with structure decomposition
python multi_structure_mesh_reader.py <input.vtk> <output.vtk> <point_data_name>

# Apply SSM deformation
python multi_structure_mesh_reader.py <input.vtk> <output.vtk> <point_data_name> \
    <mean_mesh.vtk> <components.txt> <variances.txt> <num_components>

# Apply SSM deformation with explicit shape parameters
python multi_structure_mesh_reader.py <input.vtk> <output.vtk> <point_data_name> \
    <mean_mesh.vtk> <components.txt> <variances.txt> <num_components> <shape_params.txt>

Simple Visualization

The mean shape of the Auto-FWLL-SSM, visualized in 3D Slicer with per-structure color labels:

visualize_3dslicer.py is a 3D Slicer Python script for automated mesh visualization and screen capture.

# Usage (run via 3D Slicer's built-in Python)
"<path/to/Slicer>" --python-script visualize_3dslicer.py <input.vtk> \
    --capture <output.png> --auto-crop

Options: --array, --colormap, --opacity, --range-min, --range-max, --rotx/y/z, --zoom, --pan-x/y, --parallel, --fov, --capture.

Data Availability

Under construction

Citation

Under construction