Projects

HippUnfold is a world-leading software for analysis of the hippocampus. The hippocampus is a "canary in a coal mine" brain region - a sensitive and early indicator of neurological conditions like dementia, epilepsy, and many others. It is also critical in memory and high-order cognition across all mammals. Some studies are already revealing HippUnfold's potential for translation to the clinic!

Recently, we are creating a major overhaul of HippUnfold to run more lightly, smoothly, and using best-practices including BIDS, snakemake, and git continuous integration (version 2.0). We'd love to get input and/or help with outstanding issues on the new release, and would be especially keen to get feedback on its ease-of-use for new users and its Documentation.

Come to the HippUnfold Tutorial at 1PM Wednesday for an introduction!

Registration can be tricky when there is low signal-to-noise ratio, signal dropout, and geometric distortions. This technique combines deep-learning based modality-agnostic segmentation with with conventional analytic registration methods to generate precise warpfields even in low-quality data. This type of registration can also be faster at higher resolutions, due to the simplicity of the label maps. The project will involve writing an optimizing python code, writing tests, some documentation, and assessing the quality of registration. This project is a good fit for anyone eager to learn more about python programming, registration, and convolutional neural networks, but even if you're brand new to everything, we can catch you up to speed!

Neurosurgical procedures, such as deep brain stimulation or lesion resection, depend on the precise alignment of multiple types of scans to accurately target specific structures. Currently, clinical quality control relies on visual inspection, which is insensitive to subtle yet significant misalignments. To address this, our lab developed the anatomical fiducials (AFIDs) framework comprising 32 easily identifiable anatomical landmarks to measure alignment accuracy at the millimetre scale. However, this protocol was only validated on T1-weighted (T1w) MRI scans showing anatomical information. In this project, we aim to extend and validate the AFIDs framework for diffusion MRI-derived functional anisotropy maps (FA maps), which show white matter tracks in the brain. We are looking for anyone interested in learning more about neuroanatomy to join our project apply the anatomical fiducial protocol to annotate FA maps!