MultiRecon
Machine Learning for Multimodal Medical Image Reconstruction

Key Features
-
Joint Reconstruction Models
Deep learning architectures that simultaneously reconstruct PET/CT, PET/MRI and multi-energy CT images, exploiting shared anatomical and functional information. -
Dose Reduction
Algorithms designed to maintain diagnostic image quality at lower radiation doses, contributing to safer imaging protocols.
How It Works
-
Multichannel Varational Autoencoders (VAEs)
Multichannel VAEs to generate multiple images from a single latent variable combined with iterative reconstruction algorithms, thus allowing the information to be shared between the channels. -
Multichannel Diffusion Models (DMs)
Combining iterative reconstruction algorithms with multichannel DMs to sample multiple images simultanesouly while taking into account inter-channel information.
Core Team
- Alexandre Bousse (LaTIM, Project Lead)
- B. Sixou (CREATIS)
- V. Maxim (CREATIS)
- Claude Comtat (SHFJ)
- Florent Sureau (SHFJ)
- Catherine Chez-Le Rest (CHU Poitier)
Publications
2025
- Joint Reconstruction of the Activity and the Attenuation in PET by Diffusion Posterior Sampling: a Feasibility StudyIn International Conference on Fully Three-Dimensional Image Reconstruction in Radiology and Nuclear Medicine, 2025
- MHUconnect: Multi-Head U-Net Connecting All Energy Bins For Synergistic Spectral CT ReconstructionIn International Conference on Fully Three-Dimensional Image Reconstruction in Radiology and Nuclear Medicine, 2025
- Multi-Branch Generative Models for Multichannel Imaging With an Application to PET/CT Synergistic ReconstructionIEEE Transactions on Radiation and Plasma Medical Sciences, 2025
- Joint Reconstruction of Activity and Attenuation in PET by Diffusion Posterior Sampling in Wavelet Coefficient SpacearXiv preprint arXiv:2505.18782, 2025
- Material Decomposition in Photon-Counting Computed Tomography with Diffusion Models: Comparative Study and Hybridization with Variational RegularizersarXiv preprint arXiv:2503.15383, 2025
2024
- Spectral CT Two-step and One-step Material Decomposition using Diffusion Posterior SamplingIn European Signal Processing Conference (EUSIPCO), 2024
- Synergistic PET/MR reconstruction with VAE constraintIn EUSIPCO 2024-32nd European Signal Processing Conference, 2024
- Diffusion posterior sampling for synergistic reconstruction in spectral computed tomographyIn 2024 IEEE 21st international symposium on biomedical imaging (ISBI 2024). IEEE, 2024
- Bimodal PET/MRI generative reconstruction based on VAE architecturesPhysics in Medicine & Biology, 2024
- A Review on Low-Dose Emission Tomography Post-Reconstruction Denoising With Neural Network ApproachesIEEE Transactions on Radiation and Plasma Medical Sciences, 2024
- Uconnect: Synergistic Spectral CT Reconstruction With U-Nets Connecting the Energy BinsIEEE Transactions on Radiation and Plasma Medical Sciences, 2024
- Systematic Review on Learning-based Spectral CTIEEE Transactions on Radiation and Plasma Medical Sciences, 2024
- CConnect: Synergistic Convolutional Regularization for Cartesian T2* MappingarXiv preprint arXiv:2404.18182, 2024
2023
- Joint PET/CT Reconstruction Using a Double Variational AutoencoderIn IEEE Nuclear Science Symposium, Medical Imaging Conference and Room Temperature Semiconductor Detectors Conference, 2023
- Synergistic PET/CT Reconstruction Using a Joint Generative ModelIn International Conference on Fully Three-Dimensional Image Reconstruction in Radiology and Nuclear Medicine, 2023
- VAE constrained MR guided PET reconstructionIn International Conference on Fully Three-Dimensional Image Reconstruction in Radiology and Nuclear Medicine, 2023
2022
- Multi-channel convolutional analysis operator learning for dual-energy CT reconstructionPhysics in Medicine & Biology, 2022
For more details, visit our project website or check out the full Publications list on this site.