Deep-learning models to investigate the link between Major Depressive Disorder (MDD) and cardiovascular diseases using structural brain MRI.
Lead analyst: Clara Jiang (The University of Queensland)
Evidence from observational and genetic studies has shown a strong bidirectional association between depression and cardiovascular diseases (CVD), where individuals with depression have a higher risk of developing CVDs and vice versa. Understanding the effects of commonly used CVD medications on depression is thus important to prevent unintentional adverse effects, and to explore potential drug repurposing opportunities.
The proposed study aims to use machine-learning and deep-learning algorithms to understand how brain structure can predict the risk of depression and CVD comorbidities, and leverage the depression-related neuroimaging characteristics and statistical genomic approaches to infer sex-specific causal effects of CVD medication on depression.
The findings from the study are anticipated to improve the current understanding of the mechanisms underlying the depression-CVD comorbidity, as well as to inform the potential effects of CVD medications on depression-related brain biology.
Status: ONGOING – MANUSCRIPT IN PREPARATION
Choice of processing pipelines for T1-weighted brain MRI impacts association and prediction analyses
Lead analyst: Elise Delzant (Inria Paris)
The vast amount of data from the UK Biobank, offers an unprecedented opportunity to improve robustness and reproducibility in neuroimaging. In particular, little is known about the impact of MRI processing pipelines on neuroimaging results (robustness to processing). We extensively compared 5 commonly used representations of the grey matter, which included 29 traits and several types of analyses (association, prediction). We quantified the strengths and limitations of the different pipelines, which can help researchers make more informed choices to select the MRI processing best suited to their research question. To tackle this research question, we used and repurposed several methods and software used in genetics. In particular, we used linear mixed models and optimized methods that can handle ~40,000 brain MRI, each containing 150,000-650,000 brain measurements.
Status: MANUSCRIPT SUBMITTED
Genome-wide association study of the choroid plexus volume in the UKBiobank
Lead analyst: Arya Yazdan-Panah (INRIA, Paris Brain Institute)
The Choroid Plexuses (CP) are a veil-like structure located in the lateral ventricles of the brain. They play a key role in keeping the brain homeostatic state. Multiple studies have suggested changes in CP volume (CPv) in various neurological conditions such as Alzheimer’s disease (AD), Parkinson’s disease (PD, and Multiple Sclerosis (MS). Previous studies have relied on manual segmentations of the CP, resulting in small sample sizes, or on noisy automated segmentation that also limits statistical power.
We have trained and applied to the UKBiobank (N>40,000) an automatic segmentation method for the CP from T2-weighted FLAIR MRI. We will perform Genome-Wide Association study, and a whole set of genetic analyses, to shed light on the genetic variants infludencing CPv and how they relate to diseases of the brain or immune disorders.
Status: ONGOING – MANUSCRIPT IN PREPARATION
