eardi / sm-fpca Goto Github PK

This is a MATLAB implementation of the Smooth Manifold FPCA algorithm introduced in E Lila, JAD Aston, LM Sangalli (2016). The algorithm computes the principal components, and associated scores vectors, of a collection of functions whose domain is a 2D surface represented by a triangulated surface.

Set the working directory by moving, with the command cd 'your_working_directory', to the root directory containg the SM-FPCA scripts. Run the script Init.m with the command run('Init.m'). This script compiles some parts of the library Felicity here used to compute the Finite Element discretization matrices.

The file fPCA_script.m contains a complete running example on synthetic data generated on a triangular surface representing the brain stem. If this runs correctly, it should be possible to apply the algorithm to your dataset by changing the paths pointing to the data.

The resulting scores vectors are saved in a .csv file. The resulting principal components are saved in the .vtk file format, which can be vizualized in Paraview.