Recent fMRI acceleration techniques have allowed for sub-second temporal resolution, but are still impacted by decreased SNR, inter-slice crosstalk, and eddy current artifacts. A novel acquisition and reconstruction approach, dubbed Hi-Fi fMRI, combines a spiral phyllotaxis sampling trajectory, a compressed-sensing algorithm, and application of motion correction directly to k-space to obtain minimal distortions and high SNR, as well as a 250 msec temporal resolution (chosen during image reconstruction). Here are the protocols for the experiments to activate visual cortices, auditory cortices, and the cortices for resting-state.

• Hi-Fi fMRI visual

• Hi-Fi fMRI audio

• Hi-Fi resting state

The protocols for the projects MR-Eye and MR-Eye-Track.

MR-Eye 2.0: Motion poses significant challenges to diagnostic accuracy and treatment planning in Ophthalmology. The project investigates a clinical 3-Tesla MRI protocol for eyes that achieves high spatial resolution and resilience to eye motion. To achieve this, a motion artifact correction strategy is integrated into a compressed sensing (CS) image reconstruction that leverages eye-tracking signals. This motion-robust reconstruction is demonstrated on a continuous gradient recalled echo (GRE) sequence with fat suppression (LIBRE). 

MR-Eye-Track: The project aims to introduce a novel approach relying on the hypothesis that supervised machine learning (ML) algorithms can learn from the raw MRI data the patterns associated with eye motion and specific eye positions, thereby eliminating the need for an ET to obtain 3D motion-resolved eye images. 

The dataset of Eye-Tracking trajectories of patients with spatial neglect

Patients with right hemisphere damage and visual neglect have severe problems to orient attention towards left-sided objects, often associated with the tendency to produce inappropriate rightward saccades. In its most severe form, this tendency can assume the compulsive character of a rightward deviation of gaze as soon as the visual scene deploys (so-called “magnetic attraction of gaze”). However, little is known about the exact nature of inappropriate rightward saccades, their relation with impaired conscious perception of left-sided stimuli, and their lesional correlates. To explore these issues, we studied three groups of patients with right brain damage: patients with signs of left visual neglect associated to left homonymous hemianopia, neglect patients without hemianopia, and patients without neglect or hemianopia. Participants searched for a gap missing within a target, presented among distractors. Manual responses for target detection were required, while participants were encouraged to move their eyes during search. Endogenous attention could be summoned to the target location by a central cue. 

MR-Eye atlas is a novel digital atlas constructed from MR images (T1-weighted MRI acquired at 1.5T) of a large-scale population of healthy volunteers. It gathers male, female, and combined structural atlases constructed from 594 males and 616 females, with corresponding probability maps of the different labels projected onto the average templates. The atlases include 9 regions of interest: lens, globe, optic nerve, intraconal and extraconal fat, and four rectus muscles (lateral, medial, inferior, and superior). This release also shares the pretrained nnU-Net model weights, code, and a web platform for running the segmentation pipeline and exploring the atlas in 3D.

Cardiac 2D Cine MRI Dataset is 2D CINE MRI data acquired once with a cartesian trajectory and once with a radial trajectory in four chamber view on helthy volunteers. Each data set is accompanied with an example of a reconstructed image produced with our MRI reconstruction toolbox. 

The toolbox is a Matlab/C++ MRI reconstruction software highly documented, aiming to redefine our understanding of MR imaging by guiding the experimenter to obtain high-quality reconstructions, while experimenting the maths and physics behind it.

The documentation is provided on the website.

The package is available on GitHub.

The official codes are available on GitHub for the paper: 

Machine learning algorithms on eye tracking trajectories to classify patients with spatial neglect

A-Eye web platform is an online tool for automated 3D MRI segmentation of human eye structures. Users upload a T1-weighted orbital MRI and receive fully-automated segmentation masks for the same 9 eye structures covered by the atlas, together with quantitative biomarkers — axial length and volumetry — delivered as a ready-to-use zip archive. 

The pipeline is based on nnU-Net and was trained and validated on a large-scale, multi-centre cohort of 1,245 healthy adult subjects.