Magnetic resonance-based eye tracking using deep neural networks

DeepMReye (camera-less solution) use CNN to reconstruct gaze position from MR signal of the eyeballs, see accuracy below:

Paradigms:

1. fixations

2. smooth pursuit on circular or star-shaped trajectories

3. free viewing

Main arguments:

1. eye tracking should be included in fMRI research even when viewing is not part of the central task -> insights in brain functions and behavioural readout during scanning

2. MR compatible camera eye trackers track gaze position with high temporal and spatial resolution -> but need specialized equipment

3. MR based camera-less deep learning solution based on MR eyeball signal data

papar