Neurodevelopmental and neurodegenerative disorders exhibit brain dysmorphologies and clinical symptoms that vary significantly due to genetic and environmental factors, including genetic background, sex, age, stress, and life experience. Understanding this heterogeneity is crucial for improving the diagnosis, prognosis, and treatment of these disorders.
Our research focuses on the genotype-phenotype correspondence in genetic and psychiatric disorders, such as Down syndrome, Alzheimer’s disease, schizophrenia, and bipolar disorder. We employ both traditional and advanced neuroimaging and morphometric methods (e.g., volumetric analysis, geometric morphometrics, and sulcal pits analysis) to assess brain morphological variation. However, large-scale analysis is challenging. Artificial Intelligence (AI) techniques, particularly generative and deep learning models, can greatly enhance our analysis by automatically identifying subtle, high-dimensional morphological changes, enabling the processing of large datasets, and uncovering patterns that can advance our understanding of brain disorders.
In this project, we seek access to the UK Biobank (brain MRI images, genome data, and health records) to gather a large and diverse sample of individuals. This will allow us to define patterns of brain morphological variation in a reference population, which we will then compare with our patient samples. Our goal is to identify patterns of brain dysmorphology in each diagnosis and explore their variation according to age, sex, genetic, and clinical variables. We hypothesize that by combining traditional methods with AI approaches, we can leverage their complementary strengths to discover previously unreported brain alterations. Furthermore, by correlating morphological, genetic, clinical, and cognitive data, we aim to generate multi-modal biomarkers that improve diagnostic tools. Detailed disease profiles will lead to earlier detection and better health outcomes.