The project aims are part of DBT-Wellcome Trust India Alliance Early Career Fellowship awarded to the PI.
Research Question:
Can neuroimaging biomarkers for vascular cognitive impairment (VCI) be extracted, and can specific anatomic locations in the brain be identified where the likelihood of these biomarkers increases with various clinical/demographic risk factors (e.g., high blood pressure) in an aging population, thereby validating their utility in the differential diagnosis of VCI?
Project Aims:
This project aims to develop automated, deep learning-based, open-source methods to extract maximal information from neuroimaging data (e.g., lesion count/volume) to build decision-support tools that could assist in the differential diagnosis of VCI.
Proposed methodology:
1. Extraction of Clinically Useful Neuroimaging Biomarkers and IDPs for VCI:
– Utilize semi-supervised learning methods with limited manual annotations.
– Develop privacy-protection techniques (e.g. federated learning) for multi-centre adaptation in real-time.
2. Determining the Clinical Impact of the Spatial Distribution of Biomarkers:
– Study the relationship between the spatial distribution of biomarkers and clinical factors.
– Identify regions with a high probability of biomarker co-occurrence.
– Correlate VCI IDPs with various clinical and demographic factors.