| Description |
PERFORM Centre being equipped with state of the multimodal medical imaging facilities including notably a 3T Magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET), Single Photon Emission Computed Tomography (SPECT), MR compatible high-density Electro-EncephaloGraphy (EEG) and clinical Ultra-Sound (US). Medical image pipelines usually require large computational resources for every participant, whereas common studies in neurosciences, as several research projects taking place at PERFORM, are involving large cohorts of participants. Several important metical image processing pipelines will be considered. Some pipelines are being recognized at the international level by our community (e.g. extraction of the cortical surface from anatomical MRI, analysis of functional connectivity within brain networks from resting state fMRI data, analysis of anatomical brain connections using diffusion MRI). Other pipelines of interest for PERFORM researchers are being developed locally some the supervisors of this project (estimation of connector hubs from resting state fMRI, measurement of information integration within and between brain networks).
The overall objective of this project is first to develop the environment that will allow those pipelines to be processed in an efficient and accurate manner on PERFORM local cluster composed of 200 powerful workstations. More concretely, the students will be in charge of proposing optimal parallelization strategies, benchmarking our local implementations of those pipelines, validating the results in an efficient manner, while proposing quantitative metrics to quantify these evaluations. Finally, once the processes will be fully validated within PERFORM local environment, tutorials and local training sessions will be prepared in order to provide those tools to PERFOM community of researchers.
|