Available Projects

The project aims to enhance the functionality of the existing optical mapping tool by developing AI and machine learning for image processing. It is a joint project between the School of Engineering and the Medical School at the University of Warwick. This PhD will be ideal for a candidate with a background in mathematics, physics, computer science/engineering or machine learning approaches.

Primary supervisor: Dr Igor Khovanov - Email: I.Khovanov@warwick.ac.uk

Project detail:
Cardiac optical mapping is a powerful technique for measuring electrical signals from cell monolayers to whole hearts at high spatial and temporal resolution. This method involves the infusion of cardiac preparations with voltage and/or Ca²⁺ fluorescent sensors. Once these indicators are illuminated [e.g. by light-emitting diodes (LEDs)], the resulting fluorescence is captured by high-speed cameras. Consequently, optical recordings of cardiac action potentials, Ca²⁺ transient morphology, and conduction can be obtained, providing crucial electrophysiological insights in health and disease.
Optical mapping experiments typically collect high-resolution data (>1 kHz sampling rate) from thousands of locations simultaneously. However, due to short exposure times, high spatial resolution, heterogeneous dye loading and tissue illumination, signals are often noisy and further corrupted by technical artefacts, including cardiac motion. To address these challenges, we previously developed ElectroMap, a MATLAB-based platform for the analysis of optical mapping datasets. ElectroMap is now the most widely used open-source software for optical mapping analysis, enabling researchers to generate activation, conduction, alternans, and arrhythmia maps across species and experimental conditions. Nonetheless, significant user input is still required for noise reduction, segmentation, and accurate identification of electrophysiological features.
To overcome these barriers, we are developing ElectroMap-AI: an integrated artificial intelligence extension of ElectroMap. ElectroMap-AI will harness machine learning and deep learning approaches to:
1. Improve signal quality: denoising traces and compensating for motion artefacts while preserving key features of the cardiac action potential.
2. Automate feature detection: enabling unbiased identification of activation and repolarisation times, Ca²⁺ transient dynamics, and conduction patterns.
3. Predict arrhythmia susceptibility: identifying high-dimensional patterns within optical signals that precede re-entry, alternans, or conduction slowing.
4. Enhance scalability and reproducibility: reducing reliance on manual parameter tuning, and enabling robust cross-comparison of datasets across laboratories, preparations, and species.
5. Automate image segmentation: removing background fluorescence and distinguishing anatomical regions of interest (e.g. atria vs ventricle, left vs right chambers).
Through ElectroMap-AI, we aim to reinforce and extend the central role of optical mapping in cardiac research. By combining open-source accessibility with state-of-the-art AI automation, we will provide the most comprehensive analysis platform to date, ensuring objective, scalable, and reproducible insights for both new and experienced research groups. In doing so, ElectroMap-AI will establish itself as a transformative tool in cardiac research, accelerating discovery and enabling new mechanistic and translational insights into arrhythmia and heart disease.
The project is based on research led by Dr Chris O’Shea from the Heart Rhythm Research group at the University of Warwick Medical School.