Algorithms and Methods for Imaging of Brain Activity from Non-Invasive Techniques

The imaging of brain activity, also called “Functional Neuroimaging”, is used to understand the relationship between activity in certain brain areas and specific functions. These techniques include fMRI (functional Magnetic Resonance Imaging), PET (Positron Emittance Tomography), EIT (Electrical Impedance Tomography), EEG (ElectroEncephaloGraphy) and DOT (Diffuse Optical Tomography) and are widely used in the study of brain activity. In addition to clinical usage, analysis of brain activity is gaining popularity in others recent fields, i.e. Brain Computer Interfaces (BCI) and the study of cognitive processes. In these contexts, usage of classical solutions (fMRI and PET) could be unfeasible, due to their low temporal resolution, high cost and limited portability. For these reasons, portable low cost techniques are objects of the proposed thesis's research, with focus on DOT and EEG. The main contribution of this thesis focuses on the implementation of a numerical solver for DOT based on the radiosity-diffusion model, integrating the anatomical information provided by a structural MRI.In particular, we obtained a 7x speed-up over an single run of isotropic-scattered parallel Monte Carlo engine for a domain of 2 million voxels, with an accuracy comparable to 10 runs of anisotropic scattered Monte Carlo in the same geometry. The speed-up significantly increases for larger domains, allowing one to compute the light distribution of a full human head (about 3 million voxels) in 116 seconds for the platform used. The secondary contribution of this thesis focuses on EEG and it concerns the implementation of software libraries for time-domain source localization in the scope of an open-source framework called Creamino, which can be used to simplify and speed-up the design of BCI systems. It consists of firmware and software libraries that allow designers to connect new EEG platforms to software tools for BCI.