2021

Project title: "Understanding Processes Implicated in Cognition and Age-related Cognitive Decline using Experimental Biology and Computational Analysis"

Application deadline

16th December 2021

Description

We recently discovered a novel role for the SORCS2 gene in DNA double-strand break (DSB) formation (Gospodinova et al., 2021). We showed that loss of the receptor in mice is associated with elevated DSB levels in the mouse hippocampus. In addition, knocking out SORCS2 in a human neuronal cell line led to increased DSB formation and reduced neuronal viability. These findings are of interest as DNA damage is linked to age-related cognitive decline, and transient neuronal DSBs also occur during learning and memory formation. 

The aims of the project are to determine the causes of the DSBs and the impact of SORCS2 loss on the cell. SORCS2 is implicated in neuronal activity, NMDAR trafficking and oxidative stress. The student will perform experiments to investigate whether deficits in these processes underlie the increased numbers of DSBs in the neurons lacking SORCS2. In parallel, they will investigate the impact of SORCS2 loss on the cell. For example, cell painting, which provides a phenotypic signature via the simultaneous application of fluorogenic protein stains that label multiple cellular structures and subcellular organelles, will be combined with automated high content imaging to extract ~1000 cellular features (various measures of size, shape, texture, intensity etc). These data can be used to create a machine learning model to enable an unbiased classification of cellular phenotypes in a high throughput manner. Thus the project will lead to greater understanding of functions implicated in learning and memory processes and age-related cognitive decline. 

Supervisors

The project is co-supervised by Prof Neil Carragher (who has expertise in the cell painting area) and Dr Stuart Aitken (who has expertise in machine learning).

Enquires

For enquiries, please contact Ms Kathy Evans:

Kathy.Evans@ed.ac.uk

Further details

Find more details and how to apply link and instructions on the Eastbio Doctoral Training Centre website.

PhD studentship information on Eastbio Doctoral Training Centre website

Project title: Developing explainable Machine Learning approaches for brain health prognosis and patient stratification in a nationwide dataset

Application deadline

20th January 2022

Description

Background

Cerebral small vessel disease (SVD) is estimated to affect up to 1 billion people worldwide. SVD refers to a syndrome of clinical and imaging findings that result from pathologies of the small blood vessels in the brain [1]. SVD causes up to 45% of dementia and accounts for about 20% of all strokes. The cause of the pathology is poorly understood since, until recently, it has been difficult to identify the brain abnormalities in people in vivo and there are no reliable experimental models [2]. There has been a rapid advance in understanding of SVD with the wider availability of magnetic resonance imaging (MRI). This shows overt SVD lesions such as, among others, white matter hyperintensities, lacunes, microbleeds, and perivascular spaces [3]. Although risk factors (hypertension, smoking, and diabetes) increase the risk of SVD, these together explain only a small proportion of variance in SVD lesions [1], indicating that much greater insight into the small vessel abnormalities are needed for patient stratification and prognosis. The supervisory team is world-leading in the development neuroimaging analysis approaches to quantifying SVD lesion burden in Compute Tomography (CT) and MRI scans. However, work to date has occurred primarily in the context of clinical studies and trials. There’s a pressing need to adapt these techniques to routinely collected scans, which suffer from high variability in quality, and design analysis pipelines that can be integrated into routine clinical practice. The Scottish Medical Imaging (SMI) Service, launched in November 2020, has created a national database of deidentified routine clinical images that can be linked to outcome data within the National Safe Haven. The SMI database has been created from a copy of the NHS Scotland Picture Archiving & Communication System dataset covering the years 2010-2017. This is a unique resource worldwide and has potential to lead to a step change in SVD understanding.

Aims

The main aim of the study is developing neuroimaging analysis approaches for the automated quantification of SVD lesions in routinely acquired CT/MRI scans and develop predicitive models for patient stratification and prognosis based on these interpretable features. Progress towards this aim will be delivered based on the following objectives: Page 4 of 7

1. Using the SMI service, build a virtual cohort of 5,000 patients (and the same number of matched controls) admitted to hospital with: ischemic stroke or vascular dementia with available MRI structural data (T1, T2, and FLAIR sequences) and/or CT; differentiating them from those without any pathology, neoplastic lesion(s), haemorrhagic stroke, a genetic disease, brain injury, or malformations. Link the imaging database to clinical outcomes in the National Safe Haven.

2. Develop methods for: a) automated quality scoring, b) upsampling and reconstructing (super-resolution) the scans using generative adversarial models (GANs)

3. Based on a subset of the images that have been scored visually for SVD lesions as well as other features, develop deep learning approaches to predict (a regression problem in ML): a) the quantification of individual lesion burden, and b) aggregated measures of SVD disease burden such as the Brain Health Index developed by the supervisory team [4].

4. Investigate associations between these automated measurements and patient outcomes related to brain health in the period 2010-2017. Based on these univariate associations, demographics, and brain health history, develop predicitive models for patient stratification and prognosis.

Training Outcomes

The student will receive state-of-the-art training in the core disciplines of image analysis, computational modelling, statistical methods, and data science while gaining expert knowledge in the context of brain SVD and in neuroscience more generally. This highly interdisciplinary approach is well aligned with the “T-shaped researcher” training requirements identified as key in the DTP.

Supervisors

Dr Miguel O. Bernabeu

Dr Maria del C. Valdés-Hernández

Professor Adam Waldman

Further details

Find more details and how to apply link and instructions on the Precision Medicine Doctoral Training Programme website.

PhD studentship opportunity with Precision Medicine Doctoral Training Programme (external website)