My research lies at the intersection of inverse problems, statistical reconstruction, optimization and computational imaging.
I develop probabilistic models and algorithmic methods for tomographic reconstruction, with a particular interest in low-count regimes, Monte Carlo simulations, sensitivity estimation and physically informed system models.
My PhD research was conducted under the supervision of Jérôme Idier ( LS2N — SiMS team), Simon Stute and Thomas Carlier ( CRCI$^{\text{2}}$NA — Nuclear Oncology team). It focused on statistical tomographic reconstruction for three-photon PET imaging and hybrid PET/Compton cameras, in the context of the XEMIS2 system.
More broadly, I am interested in statistical and numerical methods for ill-posed inverse problems beyond a single imaging modality.
I am interested in the mathematical and computational formulation of inverse problems arising in imaging systems. My work focuses on the construction of direct models, system operators and reconstruction algorithms adapted to complex acquisition geometries.
A central part of my research concerns statistical reconstruction methods, especially list-mode maximum-likelihood methods and EM-type algorithms. I am particularly interested in multi-class observation models, low-count data and probabilistic descriptions of the acquisition process.
I use Monte Carlo simulation to model detection processes, estimate class-dependent sensitivities and validate reconstruction pipelines. This includes the development of simulation tools and data conversion workflows for tomographic imaging systems.
My background in combinatorial and continuous optimization strongly influences my research. I am interested in exact and approximate algorithms, numerical optimization, sparse modelling and algorithmic strategies for large-scale inverse problems.