Dr Philippe Ciuciu is CEA Research Director at NeuroSpin (CEA, Université Paris-Saclay, France) where he leads the interdisciplinary research project focusing on “High Resolution fMRI”. He is also head of the Compressed Sensing group in the Inria Parietal team (Inria Saclay-Île-de-France). He has more than 20 years of experience in neuroimaging data analysis with more than 200 research outputs in the field. His current research interests are in developing accelerated acquisition and image reconstruction techniques for magnetic resonance imaging (MRI) with applications in clinical and cognitive neuroscience at 3 and 7 Tesla.
As IEEE Senior Member, Philippe has been involved in the Bio-Imaging Signal Processing (BISP) technical committee of the IEEE International Symposium in Biomedical Imaging for 6 years (2013-2018) and then has become representative of the IEEE Signal Processing Society in ISBI (2019-2020). He recently joined the editorial board of the IEEE open journal of signal processing as senior area editor. Philippe is also vice chair of the Biomedical Image and Signal Analytics (BISA) technical committee of the EURASIP society.
He will co-organize a summer school funded by Institut Pascal (Université Paris-Saclay) in June 2021 that will focus on “Interactions between AI and Signal&Image processing”.
Recent advances in acquisition and reconstruction for accelerated MRI
In the last decade, the application of Compressed Sensing (CS) theory to MRI has received considerable interest and led to major improvements in terms of accelerating data acquisition without degrading image quality. These achievements have mostly relied on (i) massive under-sampling of existing k-space trajectories (either Cartesian lines or non-Cartesian radial and spiral patterns) in 2D and 3D imaging and (ii) nonlinear reconstruction algorithms that promote either sparsity of the reconstructed MR image using a fixed dictionary or transform, or a low-rank representation in an appropriate domain. Two recent complementary research directions are starting to supplant this classical CS setting: first, the design of optimization- or learning-based under-sampling schemes and second, the advent of machine learning tools (e.g., dictionary and deep learning) for MR image reconstruction.
The course focuses on these new trends in CS MR acquisition and image reconstruction and is split in two parts accordingly.
The course is organized as follows:
- Part I: Basics of MR physics
- Part II: From traditional CS in MRI to optimization and learning driven accelerated acquisition
- Part III: Sparsity-promoting CS reconstruction
- Part IV: Machine Learning for MR image reconstruction
To make the course even more appealing and provide concrete illustrations on traditional and advanced approaches, a special hands-on session is proposed during the last 2 hours. This tour will rely in particular on our PySAP-MRI software package for MR image reconstruction and Jupyter notebooks available in PySAP-tutorials. The latter can be run either on Binder or GoogleColab, so without any specific installation or operating system requirements.