Femur Modeling

Statistical shape analysis of femur bones using PCA, neural networks and diffeomorphic registration (LDDMM).

Overview

This project implements a pipeline for 3D femur shape modeling:

• Custom C++ neural network library with linear algebra operations (using Eigen)
• PCA-based dimensionality reduction for shape representation
• LDDMM registration for computing diffeomorphic mappings between femur shapes
• Python bindings via pybind11 for integration with visualization tools

Building the C++ Code

Requirements

• CMake >= 3.10
• C++17 compiler
• Eigen (included in lib/eigen-5.0.0/)

Build Instructions

mkdir -p build && cd build
cmake ..
make

Executables are generated in the bin/ directory.

Running Tests

cd build
ctest

Documentation (optional)

make doc        # Generate Doxygen documentation

Additional Documentation

• LDDMM Pipeline: See lddmm/README.md
• Visualization Tools: See visualization/