< Back to previous page

Project

Learning-based computational strategies for multimodal image analysis of neurovascular diseases

Neurovascular diseases such as ischemic stroke are one of the most prevalent causes of disability worldwide. In ischemic stroke, blood flow to a specific region of the brain is hampered by a thrombosis or embolism occluding a cerebral artery. The main treatment strategies of ischemic stroke consist of restoring perfusion through intravenous injection of a thrombolytic drug and/or endovascular thrombectomy, with time from onset to reperfusion being a determining factor in terms of patient outcome. The severity of the stroke is clinically assessed based on conventional perfusion parameters such as cerebral blood flow and time-to-maximum extracted from a 4D perfusion CT scan acquired prior to treatment. But the outcome thereof is affected by image quality, by computational factors and by the lack of inclusion of treatment specific parameters. The aim of this thesis is to investigate alternative, learning-based strategies that allow for a more comprehensive analysis of the patient imaging data to more reliably predict treatment outcome in stroke and other neurovascular diseases by accounting as well for imaging data of previous cases, non-imaging data of the patient (e.g. EEG) and metadata in the form of clinical parameters.

Date:18 Nov 2021 →  Today
Keywords:Biomedical Engineering, Medical Imaging, Signal Processing, Neuroimaging, Artificial Intelligence
Disciplines:Medical imaging and therapy not elsewhere classified, Image processing, Signal processing, Neurosciences not elsewhere classified, Artificial intelligence not elsewhere classified
Project type:PhD project