Projects

Diffusion MRI Transformer with a Diffusion Space Rotary Positional Embedding (D-RoPE)

Novel transformer model that combines novel absolute and relative positional embeddings to encode the geometric information of the different volumes in a diffusion MRI sequence to obtain robust and translatable representation. The proposed model is able to handle acquisitions with multiple b-values and an arbitrary number of directions. 

Code  | Paper

DTI Processing Pipeline: A lightweight, end-to-end workflow for diffusion MRI preprocessing and quality control

This pipeline provides distortion ocrrection, within-session alignment/merge, registration to MNI, and diffusion tensor fitting (DIPY). The pipeline generates clear QC visualizations and summary tables. The design is config-driven, scalable across cohorts, and interoperable with common neuroimaging tools and AI pipelines.

Code | Website

Inferring microstructure distributions from multimodal MRI

The Panning and Translational Invariant Incremental Principal Component Pursuit (incPCP-PTI) algorithm is a fully incremental method for video background modeling that extends incremental PCP in order to cope with moving and panning cameras. 

Code | Paper

Projection into mixed-norm balls

Root-finding based algorithm for computing projections onto the l∞,1-ball. This method proposes an initial guess of the root of a search function that can be obtained with low computational cost.

Code | Paper

Panning and Translational Invariant Incremental Principal Component Pursuit

The Panning and Translational Invariant Incremental Principal Component Pursuit (incPCP-PTI) algorithm is a fully incremental method for video background modeling  that extends incremental PCP in order to cope with moving and panning cameras. 

Code | Paper