Research

Neuroimaging data acquisition and processing Lab

Our main interests consist in the development of techniques for the acquisition and processing of neuroimaging data. These include magnetic resonance imaging (MRI) and positron emission tomography (PET) but also electroencephalography and physiological signals such as actimetry and occulometry.  The curation and management of neuroimaging data is also a focus of our activities.

Magnetic Resonance Imaging (MRI) 

Researchers involved: 

MRI data are at the chore of our activities within the Cyclotron Research Centre. Thanks to its 2 onsite MRI scanners, 3T Prisma & 7T Terra (Siemens), most of the data we deal with are acquired locally. 

The MRI Physics team includes Laurent Lamalle, Mikhail Zubkov and Evgenios Kornaropoulos. They focus on the development and optimization of advanced MRI acquisition sequences on both 3T and 7T machines, such as quantitative MRI (qMRI), Chemical Exchange Saturation Transfer (CEST), Quantitative Susceptibility Mapping (QSM) as well as specific function and diffusion-weighted MRI sequences. 

Research on the MRI data processing covers a broad field of topics: 

  • Quantitative maps reconstruction for qMRI, CEST and QSM, plus their spatial processing and statistical analysis; 
  • Modelling of diffusion weighted MRI, using ‘tensor models’, “Neurite Orientation Dispersion and Density Imaging” (aka. NODDI), and tractography; 
  • Resting state fMRI and functional connectivity. 

These topics are mostly covered by Mohamed Bahri, Gregory Hammad, Evgenios Kornaropoulos, Christophe Phillips 

Positron Emission Tomography (PET) 

Researchers involved: 

The lab team has acquired experience in modelling dynamic PET data to obtain quantitative maps. Recently we focused on the use of an “image-derived input function” (IDIF), instead of arterial plasma time-activity curve (AIF) to obtain quantitative maps, which is much less invasive for the subjects scanned. We are currently developing to automatize the extraction of this IDIF from dynamic PET images.  

Electroencephalography & physiological signals 

Researchers involved: 

Electroencepahalography (EEG) & brain stimulation 

The lab has a long standing experience with polysomnographic EEG data processing, from multichannel data visualization to manual scoring and feature extraction. We also contributed to EEG source reconstruction.  

We have experience with EEG-TMS recording and modelling. Evoker-response EEG (ERP-EEG) signal recorded during “transcranial magnetic stimulation” (TMS) combined “dynamic causal modelling” (DCM) with cortical column “neural mass model” (NMM-DCM) provides a computational way to look at neuronal excitability and its modulation. 

Recently we worked on accurate electro-magnetic head modelling for application such as EEG forward problem solution and “trans-cranial direct current stimulation” (tDCS). The aims of the project for tDCS are to 1/ better understand how tDCS can affect cognitive performances and 2/ optimize the experimental setup to maximize the cognitive effect. 

Physiological signal analysis 

We have developed a software toolbox for actigraphic data. This tool allows the systematic analysis and modelling of large scale, i.e. month long and 100+ subjects, actigraphy data.  

Similarly we have used aye-tracking systems, typically during functional MRI acquisition, to record pupil movements and dilation over time. These occulometric data can provide insight about one’s level of wakefulness and emotional reactiveness. 

Statistics, modelling & inference methods 

Researchers involved: 

The Development in neuroimaging data acquisition and modeling lab is overall supporting the analysis of all the neuroimaging data (all flavours of MRI, PET, EEG, actigraphy,…) used other by the researchers. We have developed some expertise in data processing like 

  • image spatial (pre-)processing, such as realignment & coregistration, segmentation including lesioned brain, normalization in standard space, etc., as well as tensor fitting and tractography;
  • “statistical parametric mapping” for (massively) univariate classic inference on population(s) of subjects data; 
  • multivariate & multimodal SPM (mSPM) analysis. This allows the combination of multiple modalities in a single “general linear model (GLM) and  inference; 
  • multivariate & multimodal pattern recognition for predictive models. The “machine learning” and inference techniques were gathered in the open source “Pattern Recognition for Neuroimagin Toolbox” (PRoNTo, http://www.mlnl.cs.ucl.ac.uk/pronto/index.html )

Data curation and IT 

Researchers involved: 

The data being complex and highly valuable, we are also interested in their safe storage, complete description and clear organization. We are thus concerned by data curation and standardisation, which ensure their easier processing and sharing. This lead to a few contributions to the “Brain Imaging Data Structure” (BIDS, https://bids.neuroimaging.io/) initiative for qMRI, PET and EEG data.  

We also have some experience with data sharing across (international) research networks, while respecting GDPR rules. This typically involves setting up specific servers, for example XNAT(https://www.xnat.org ).

updated on 10/1/25

Share this page

cookieImage