Analysing Neuroimaging (MRI, fMRI, EEG, ...) Data

GIF references: All screenshot/screen-capture images taken and generated by H Muzart via the software cited (i.e. the data visualisation rendering). All data manipulation by H Muzart, showing graphical representations of the neural signals and ROIs. The software tools were developed by other people, and include ones for MRI/fMRI (SPM/SPM-12 & xjview, 3D Slicer, AFNI, TVB, nilearn modules in Jupyter notebook), EEG (Brainstorm, SPM, EEGLab), etc. None of the original human brain data was collected by me, but rather by other third-party university and clinical labs around the world (mainly UK/USA) in various experimental studies over the last 20 years. The shown brain scans and data datasets are all publicly open access, and contain no personally identifiable information.

sMRI and fMRI data analysis

Structural Magnetic Resonance Imaging (since the 1970s) and fMRI (since the 1990s), in terms of the hardware and software technology, has progressed tremendously, giving better spatial and temporal resolution, and correcting for artefacts. MRI technology has been used in non-humans and humans. For humans, it's applicable to all parts of the human body; and since then, there has been wide applications, including in cognitive sciences. I have myself been in an MRI machine for clinical purposes and science experiments, and reviewed a lot of the literature on fMRI studies, while on my UCL BSc and KCL MSc (pertaining to the engineering physics, neurophysiology, applications in cognitive neuroscience, logical procedures of participant interactions, etc).

During 2016-2019, as shown in the images below, I independently (independently from my university / academic theses) used open-source data (e.g. from openfMRI, openNeuro.org/, others) with tools like (3D-)Slicer, using nilearn in iPython/Jupyter, AFNI, BrainVoyager, 'FSLeyes' (see fsl.fmrib.ox.ac.uk/fsl/fslwiki ) and SPM-8/SPM-12 (see fil.ion.ucl.ac.uk/spm ). I also used tools like FreeSurfer and SPM for computational neuroanatomical morphometrics. The algorithms for neuroimaging data processing (MRI, fMRI, PET, EEG, ...) that constitute the 'Statistical Parametric Mapping' mathematical techniques (Friston et al 1994, 2007) are built in the software of the same name, and which runs on Matlab [link]. Overall, I have a lot more experience with spatio-temporal volumetric time-series signal processing in SPM, especially as my KCL Thesis (2020) focused on that, using data pre-acquired from the KCL IOP CNS and stored on their NaN system (that I processed via Unix/Linux). While the standard MRI and fMRI is of interest to me, there is also diffusion tensor imaging tractography (DTI/DTT), and separately (resting-state) rs-fMRI effective connectivity analyses. (Also see BioNeuroTech.com/nhanp). I am looking to analyse more of these.

My interest is on the application of fMRI-related technology to cognitive neuroscience; as well as the development of better software-based analyses (e.g. Bayesian optimisation during data collection, AI-driven visual pattern recognition) and hardware tools (cheaper high Tesla machines, larger holes for patients, accommodation of paradigms that include arm movements and simultaneous EEG/TMS/tDCS, 'portable' MRI machines, etc). I am also interested in real-time neurofeedback and pharmacological ph-MRI applications.

EEG data analysis

Electroencephalography: Electrical properties of neural systems offer the ability to look into physiology at the neuronal level (e.g. CognTech Neuronal 1 2 ) and at the whole brain level (with Electroencephalography). I met EEG theory back in the mid-2000s, but it's not until more recently in 2020, at KCL, that I got more understanding in it. I do have some separate practical experience with EEG set-up workshops and being a participant myself (at occasional times 2017-2019), having had my brain experimented on, and I have some basic experience with the set-up and issues that come from it. I also have an interest in EEG in particular now with modern-day BMIs being cheaper and more versatile, with millisecond-level accuracy, although their spatial resolution remains an issue. During 2017-2020, much data I have manipulated (not related to my university works) using Matlab and variable Runtimes on Windows (Brainstorm, EEGLab, SPM EEG), and R Studio, using open-source datasets (event-related potentials (ERPs), qEEG Oscillation bands), from [ e.g. sccn.ucsd.edu ]. I haven't yet had a chance to go into it analysing much though, only really displaying the data and interpreting it visually by eye.

MEG data analysis

Magnetoencephalography: This has many good spatial and temporal resolution promises. There are a variety of hardware systems (such as fixed-head scanner, and other large static set-ups that allow for head/body movements). As above, there are Matlab/Python toolboxes, SPM, and Brainstorm, that allows for the analysis of these. Here are some open data that I had used: fieldtriptoolbox.org/faq/open_data , datasetsearch.research.google.com.

In General, Multi-Modal, and Other

see my (empty) projects repos:

https://neurovault.org/collections/12677/

https://openneuro.org/search?mydatasets

here are some of my social forum profiles:

https://neuroimage.usc.edu/forums/u/harrym2018/summary

https://uk.mathworks.com/matlabcentral/profile/authors/9089185-harry-muzart

https://discourse.slicer.org/u/Harry-Muzart/summary

https://forum.humanbrainproject.eu/users/harry_muzart/activity

Here are the closest datasets (MRI/fMRI, MEG, EEG) from others that I could find for my investigations with regards to these cogntive paradigms:

--- visual, visuo-spatial perception

--- recognition and pattern completion

--- emotional episodic memory

--- emotional associative learning

--- human faces, emotional face, social contexts

--- spatial navigation and reinforcement learning in 3D virtual environments and in real environments

--- 1st-person POV navigation/movement

--- language processing (text + auditory, with other)

--- choice-making decision-making

[see https://docs.google.com/document/d/1aovRWdVYgTbMCUAl8VC7kr97nFnVOu2qMW153Ul8ydM/ ], also shown below.

-----------------------

I am also interested in helping develop the recording/collecting of data:

• In: fMRI; EEG; combined simultaneous fMRI-EEG; combined fMRI-tDCS-TMS; fMRI-VR; MEG-VR, EEG-VR. (in terms of the hardware technological advances, and the software, and the person/patient psychological experience; as well as paradigm administration techniques like Bayesian optimisation)

-----------------------

I am also interested in:

- CAT ('Computed Axial' Tomography), although this is an older technology.

- PET (Positron Emission Tomography), despite its higher costs and radioactive isotopes.

- Electrical impedance (EIT) data.

- ECoG - Electrocorticogram data.

- fNIRS (functional Near-Infrared Spectroscopy), which, like an EEG cap set-up, is fairly versatile.

- MRS (Magnetic Resonance Spectroscopy) for brains (like MRI brain scans and NMR chemical compositions).

- Other neural imaging, but not at the macroscopic level, rather at the neuronal level.