About Me

Pierre-Antoine Comby is currently a Post-Doc Fellow in the MIND Team (CEA/Inria). He got a PhD in Physics and Medical Imaging from Paris-Saclay University under the supervision of Dr. Philippe Ciuciu and Dr. Alexandre Vignaud in 2025. He received his M.Sc. in Signal Processing from the Ecole Normale Supérieure Paris-Saclay in 2021. His research interests include signal processing, machine learning, and computational neuroscience. He is particularly interested in the development of new methods for functional MRI reconstruction, with a focus on high-resolution imaging and accelerated acquisition.

Pierre-Antoine Comby published in several international conferences, including the IEEE International Symposium on Biomedical Imaging (ISBI), the International Society for Magnetic Resonance in Medicine (ISMRM). He has been a reviewer for the IEEE Transactions on Medical Imaging, the IEEE Transactions on Computational Imaging, as well as for the Magnetic Resonance in Medicine journal.

Functional MRI is limited by a trade-off between spatial resolution, temporal resolution, signal-to-noise ratio and brain coverage. Ultra-high-field scanners could shift that trade-off, but only if acquisition and reconstruction are designed together, and only if the resulting methods are robust and easy enough for neuroscientists to use in practice. That is what I work toward: first at 7 T during my PhD, now at 11.7 T at Neurospin.

• Simulation SNAKE is a realistic fMRI simulator that goes from neural activation and the BOLD response to raw k-space data (Imaging Neuroscience, 2025). It provides ground truth for comparing acquisition and reconstruction strategies before any scanner time is spent, and it received 2nd place in the ISMRM 2024 Reproducible Research Study Group award.
• Reconstruction Compressed sensing and learned priors for highly accelerated MRI: low-rank denoising of 7 T fMRI (ISBI 2023) and Plug-and-Play reconstruction for 3D non-Cartesian data (ISBI 2025, EUSIPCO 2025).
• Acquisition Non-Cartesian k-space trajectories turned into hardware-compliant gradient waveforms with convex optimization, and exported to the scanner through Pulseq.
• Software I am the lead developer of MRI-NUFFT (JOSS, 2025), a single interface to CPU and GPU non-uniform FFT libraries with the MRI forward model built in. I presented it in an ISMRM 2026 tutorial on open-source reconstruction software. Both MRI-NUFFT and SNAKE are used by groups outside our own team.

I foster for reproducible science that should be easily reused and abused: tested, documented, packaged, and designed so that other people can build on it. Most of my methods are published together with the software that implements them, and I regularly present that software to its users at workshops and tutorials.

I have peer reviewed several submissions for the following Journals, among others, full list on ORCID

• IEEE Transactions on Medical Imaging (TMI) (Distinguished Reviewer, 2024)
• IEEE Transactions on Computational Imaging (TCI)
• Magnetic Resonance in Medicine (MRM)