Welcome to shamo’s documentation!

Introduction

Constructing accurate subject specific head model is of main interest in the fields of source imaging (EEG/MEG) and brain stimulation (tDCS/tMS). shamo is an open source python package to calculate EEG leadfields, current flows, and electric potential distribution in the head. From a labelled 3D image of the head, the whole process is fully automatized, relying only on a few parameter files, e.g. conductivities (including white matter anisotropy) plus source and electrode locations. Since there is no non-invasive method to measure the electromagnetic (EM) properties of the head tissues, shamo can also be used to assess the sensitivity of the EM head model to these parameters.

Philosophy

The idea leading the development of shamo is to provide a versatile, intuitive and extendable toolbox for electromagnetic modelling of the head. Every object is though to be savable/loadable as a dictionary and stored as a JSON file on disk. shamo is built around three main concepts:

  1. Problem: The definition of a task to perform. Computing the EEG leadfield or simulating tDCS for examples.

  2. Solution: The object resulting from the resolution of a problem.

  3. Surrogate: If the problem-solution pair is parametric, e.g. some parameters are random variables, surrogate can be used to produce parametric models.

One of the leading rules while working on shamo was to use already existing quality tools to perform key steps. Thus, the finite element generation is achieved by interfacing with CGAL and Gmsh, the physical problem resolution is done with GetDP, the Gaussian processes are generated with scikit-learn and the sensitivity analysis uses SALib.