Thesis title: Deep Learning Techniques for Clinical Diagnosis
Medical images play an important role in medical diagnosis and treatment. Oncologists analyze images to determine the different characteristics of the deadly diseases, to plan the therapy and to observe the evolution of the disease. The objective of this thesis is to propose efficient methods for detection of two deadliest diseases, i.e. COVID-19 and brain tumors. As concerns COVID-19, it will be detected using Chest X-ray scans, while brain tumors will be identified starting from Magnetic Resonance (MR) images performing suitable segmentation procedures. The latest technical literature concerning radiographic CT images of COVID-19 shows that deep learning methods can be implemented to extract specific features of COVID-19, aiding the clinical diagnosis. For this reason, most data scientists and AI researchers work on Machine Learning methods COVID-19 for designing automatic screening procedures. Indeed, an automated method would result in quicker segmentation findings and results that would not differ as much between hospitals with various resources, resulting in a more consistent identification of brain tumors and COVID19. To improve the performance of segmentation new architectures are proposed and tested in this thesis. We propose deep neural networks for detection of COVID-19, trained on the x-ray images of patients’ lungs. Proposed architecture are based on convolutional neural networks and inception modules for brain tumors segmentation. A comparison of these proposed architectures with the baseline reference ones shows very interesting results.