Dissertation submitted in partial fulfilment of the requirements for the Degree of Master of Science in Geospatial Technologies Neurodegenerative diseases comprise a group of chronic and irreversible conditions characterized by the progressive degeneration of the structure and function of the central nervous system. The detection and classification of patients according to the underlying disease are crucial for developing oriented treatments and enriching prognosis. In this context, Magnetic resonance imaging (MRI) data can provide meaningful insights into neurodegeneration by detecting the physiological manifestations in the brain caused by the disease processes. One field of extensive clinical use of MRI is the accurate and automated classification of neurodegenerative disorders. Most studies distinguish patients from healthy subjects or stages within the same disease. Such distinction does not mirror clinical practice, as a patient may not show all symptoms, especially if the disease is in an early stage, or show, due to comorbidities, other symptoms as well. Likewise, automated classifiers are partly suited for medical diagnosis since they cannot produce probabilistic predictions nor account for uncertainty. Also, existent studies ignore the spatial heterogeneity of the brain alterations caused by neurodegenerative processes. The spatial configuration of the neuronal loss is a characteristic hallmark for each disorder. To fill these gaps, this thesis aims to develop a classification technique that incorporates uncertainty and spatial information for distinguishing four neurodegenerative diseases, Alzheimer’s disease, Mild cognitive impairment, Parkinson’s disease and Multiple Sclerosis, and healthy subjects. This technique will produce automated, contingent, and accurate predictions to support clinical diagnosis. To quantify prediction uncertainty and improve classification accuracy, this study introduces a Bayesian neural network with a spatially informed input. A convolutional neural network (CNN) is developed to identify a neurodegenerative condition based on T1weighted MRI scans from patients and healthy controls. Bayesian inference is incorporated into the CNN to measure uncertainty and produce probabilistic predictions. Also, a spatially informed MRI scan is added to the CNN to improve feature detection and classification accuracy. The Spatially informed Bayesian Neural Network (SBNN) proposed in this work demonstrates that classification accuracy can be increased up to 25% by including the spatially informed MRI scan. Furthermore, the SBNN provides robust probabilistic diagnosis that resembles clinical decision-making and accounts for atypical, numerous, and early presentations of neurodegenerative disorders.