UCL Centre for Medical Image Computing


Panagiotis Barmpoutis - Lizzie Powell - CMIC/WEISS joint seminar series

20 May 2020, 1:00 pm–2:00 pm

Panagiotis Barmpoutis - Lizzie Powell - talks as part of the CMIC/WEISS joint seminar series.

Event Information

Open to




Lizzie Powell


Title: Imaging blood-brain barrier water exchange


Abstract: The blood-brain barrier (BBB) separates the vasculature from the brain. Its main function is to regulate the transport of substances from the blood into the brain, and so plays a vital role in maintaining normal brain function. Disruption to the BBB - where the barrier becomes leaky and allows potentially toxic substances through - is associated with many conditions, including stroke, tumours and neurodegenerative diseases. Unfortunately, current non-invasive imaging methods for quantifying BBB breakdown can be unreliable, particularly for early stage, subtle damage.


In this talk, I will describe a new approach for assessing subtle BBB dysfunction, in which double diffusion encoded MRI is implemented to measure water exchange between the vasculature and brain tissue. Using simulations and a simplified two-compartment model, I will discuss the extent to which we can resolve different exchange rates across the BBB, considering the impact of scanner hardware, sequence optimisation and SNR on the accuracy and precision of model parameters.



Panagiotis Barmpoutis


Title: Modelling of multidimensional signals as third order tensor structures: Multi-lead ECG signal analysis for myocardial infarction detection and localization.


Abstract: In the first part of my presentation, I will give a brief overview of modelling of multidimensional signals as third order tensor structures and through a higher-order linear dynamical systems analysis. In the second part, I will focus on an approach that reshapes the electrocardiogram signals into a third-order tensor structure and subsequently extracts feature representations in both Euclidean and Grassmannian space, aiming to detect and localize myocardial infarction. Electrocardiogram is commonly used as a diagnostic tool for the monitoring of cardiac health and the detection of possible heart diseases. However, the procedure followed for the diagnosis of heart abnormalities is time-consuming and prone to human errors. Thus, the development of computer-aided techniques for the automatic analysis of electrocardiogram signals is of vital importance for the diagnosis and prevention of heart diseases. Experimental results of the proposed method demonstrate better the inter-correlations between signals of different ECG leads by extracting feature representations that lie in different geometrical spaces and contain complementary information with regards to the dynamics of signals.