Application of Machine Learning Techniques to Brain Magnetic Resonance Imaging in Hypertensive Patients

Hypertension is one of the main risk factors for vascular dementia and Alzheimer’s disease. To predict the onset of these diseases, it is necessary to develop tools to detect the early effects of vascular risk factors on the brain. In this thesis we will tackle the problem of hypertensive brain organ damage characterization approaching it at multiple scales and leveraging multiple techniques. The first part of the thesis will be focused on a deep learning system to perform automatic segmentation of White Matter Hyperintensities (WMH), one of the most common form of macrostructural vascular injury in the brain, on T2-FLAIR imaging. To this aim we will leverage a public dataset and compare our results with the ones achieved in the MICCAI WMH segmentation challenge. The second part of the thesis is focused on the setup of analysis pipelines for Diffusion Tensor Imaging (DTI) and resting state functional MRI (rs-fMRI), with the aim of characterize the microstructural integrity and the functional connectivity. These pipelines have been implemented on hypertensive brains to characterize the subtle brain functional and microstructural damage associated with the hypertensive condition. Finally, both approaches have been implemented in a ongoing research program at IRCCS Neuromed in the context of the heart and brain clinical research, achieving the injury characterization for the first two recruited patients of the study and field-testing the proposed brain injury characterization framework.