Projects | MANIFOLD Lab

AutoML

Mon, 01 Jan 0001 00:00:00 +0000

Using AutoML-Multiverse, we search through pipelines in an unbiased and data-driven manner to select the opimal machine learning pipeline for the task at hand.

BARCODE

Mon, 01 Jan 0001 00:00:00 +0000

ENIGMA Brain Age

Mon, 01 Jan 0001 00:00:00 +0000

Interpretable Dementia Prediction

Mon, 01 Jan 0001 00:00:00 +0000

Developing a generalisable and interpretable framework for dementia prediction.

[1] Rob Matherson, “Model predicts cognitive decline due to Alzheimer’s, up to two years out” MIT News Office (2019) https://news.mit.edu/2019/model-predicts-alzheimers-decline-0802

Low-field MRI

Mon, 01 Jan 0001 00:00:00 +0000

Motor Neurone Disease

Mon, 01 Jan 0001 00:00:00 +0000

Motor neurone disease (MND) is a rare and incurable neurodegenerative condition which affects the motor neurons in the brain, leading to muscle weakness, speech difficulties, and respiratory failure. Predicting the prognosis of MND at the diagnostic appointment would help with more efficient and effective clinical care management as well as improving clinical trial recruitment. Clinical data recorded at MND-care appointments includes ratings of the patient’s functions and neurodegeneration-related markers in the blood. Brain and spinal imaging (usually MRI) is also taken at the diagnostic appointment to rule out mimic diseases.

We are combining different sources of MND data to predict disease progression through the application of deep learning data fusion techniques. Additionally, we are exploring the relationship between the brain age paradigm and MND disease progession through the use of deep generative modelling. In collaboration with Professor Andrea Malaspina and the UCL Queen Square Motor Neurone Disease Centre https://www.ucl.ac.uk/ion/research/research-centres/motor-neuron-disease-centre

Multimodal Brain Age

Mon, 01 Jan 0001 00:00:00 +0000

Neurodevelopment

Mon, 01 Jan 0001 00:00:00 +0000

Using machine and deep learning approaches, this project aims to determine whether characteristics of brain structure (as indexed by neuroimaging) measured in infancy and early childhood can help predict poor educational outcomes in later childhood.

Normative Modelling

Mon, 01 Jan 0001 00:00:00 +0000

We are applying normative models to dementia patient data, to map the individualised neurobiological fingerprints.

Scottish Medical Imaging Archive

Mon, 01 Jan 0001 00:00:00 +0000

In collaboration with researchers from The University of Edinburgh and eDRIS (Public Health Scotland).