Research

Machine learning to enhance clinical MRI

Image Quality Transfer (IQT) can enhance medical images (MRI) by leveraging the latest advances in machine learning (random forests [3, 4] and deep learning (CNNs) [5]).

Recent advances in research MRI scanners produce stunning details (Human Connectome Project), however clinical scanners will not profit from this technological advance due to prohibitive purchase and running costs -- even for the next decade. IQT bridges this gap by learning the fine detail available in high-quality research data that is rare and expensive and by propagating it to enhance clinical data through machine learning.

See video description on YouTube for more details.

View Fibres Interactively

Explore white matter fibres from diffusion MRI.

This is a three.js and javascript proof-of-concept program to visualise dMRI fibre tracts interactively online. Also shown are three slices (axial, coronal and sagittal), with crossing fibre orientations. This demo contains about 100 fibres.

A movie to showcase some of my work in dMRI

This video displays a flyby of an axial slice of the brain. First we see the diffusion tensors from a DTI estimation. This is replaced by EAPs (ensemble average propagators) computed from 4th order tensors (modified GDTI model [1]). Finally the EAPs are replaced by their maxima (automatic maxima computation from a polynomial approach [2]) that makes fibre crossings easy to identify. Best viewed in HD with annotations.