L'invité du mois : Philippe Ciuciu

Advancing Brain Imaging: Exploring New Frontiers with AI

MRI enables non-invasive, high-precision brain imaging at 3 or 7 Tesla, but higher resolution typically means longer scan times. Traditional acceleration methods like parallel imaging are limited by signal-to-noise issues, and compressive sampling by complex, nonlinear reconstruction. Deep learning approaches improve image quality but can cause instabilities or hallucinations, requiring large supervised datasets (with ground truth) and matching acquisition schemes. Self-supervised methods exist but are ineffective for highly accelerated data.

My team’s recent work on Plug-and-Play (PnP) methods addresses these issues. Although standard PnP algorithms often trade off stability and quality, combining three key strategies (unsupervised noise-free training, preconditioning for faster optimization, and damping schemes) enables millimetric-resolution T1-weighted anatomical MRI scans in just 1 minute each for acquisition and reconstruction, outperforming costly commercial solutions. These advancements can also be applied to functional brain imaging.

Philippe Ciuciu 
CEA Fellow, CEA Research Director, Head of the Inria MIND team

CEA/NeuroSpin, Gif-sur-Yvette, France
Inria MIND team, Palaiseau, France
philippe [*] ciuciucea [*] fr (philippe[dot]ciuciu[at]cea[dot]fr)

More about Dr. Philippe Ciuciu:
 

Dr. Philippe Ciuciu earned his PhD in Electrical Engineering in 2000 and his Habilitation to Conduct Research in 2008, both from the University of Paris-Sud (Orsay, France). He began his career as a postdoctoral fellow at the French Alternative Energies and Atomic Energy Commission (CEA) in 2000. In 2012-2013, he was a guest professor at the University of Toulouse’s Department of Applied Mathematics.

Currently, Dr. Ciuciu is a CEA Fellow and Research Director at NeuroSpin (CEA Paris-Saclay), the largest ultra-high field MRI center dedicated to cognitive and clinical neuroscience. He also holds a joint appointment at Inria, the French Institute for Research in Digital Science and Technology. Since 2018, he has led the Compressed Sensing team within the former Inria-CEA Parietal team, and in 2022, he became the head of the Inria-CEA MIND (Models and Inference for Neuroimaging Data) lab, a cutting-edge research unit with 40 researchers and staff.
His interdisciplinary research spans accelerated MRI, advanced signal processing for functional brain data analysis (fMRI, M/EEG), and machine learning applications in neuroscience. Dr. Ciuciu has authored over 230 peer-reviewed publications, 6 MRI-related patents, and more than 75 articles in leading international journals.

In addition to his research, he served as Senior Area Editor for the IEEE Open Journal of Signal Processing (2019-2024) and Associate Editor for Frontiers in Neuroscience, Brain Imaging Methods (2020-2024). In 2022, he joined the editorial board of IEEE Transactions on Medical Imaging. He has played key roles in international scientific communities, serving as Vice Chair (2019-2021) of the EURASIP Biomedical Image and Signal Analytics (BISA) technical committee, a conference program committee member for ESMRMB 2021 and 2023, and a steering committee member for the International Symposium on Biomedical Imaging (2020-2022), where he represented the IEEE Signal Processing Society.

Since November 2024, Dr. Ciuciu has been leading the BrainSync at-risk research project as part of CEA’s Audace program. This initiative aims to develop foundation models of brain learning mechanisms for therapeutic innovation, leveraging AI and neuroprostheses.