MyCardium AI is developing a next generation, fully automated, real-time, platform-agnostic system for analysing cardiac MR images, enabling clinicians and researchers to increase analysis speed, reduce costs, reduce expert input time and importantly produce test results with precision and prognostic ability exceeding human experts.
Measuring the heart’s squeezing ability and size (ejection fraction, EF, volumes, mass) guides treatment for millions globally. These measurements gatekeeper the diagnosis and management of heart failure, cancer therapy heart injury, and are key drug development endpoints. The measurements are best acquired using cardiac magnetic resonance (CMR) imaging, but human analysis of the images is costly and has inherent variation. This variation creates uncertainty in the results which triggers higher levels of testing and shorter patient follow-up intervals. In some cases, it may also cause chemotherapy to be inappropriately stopped or defibrillators to be implanted in the wrong patients. Improved measurement of heart function will better improve clinical patient outcomes and save money.
A consortium of Barts Health NHS Trust (the world’s largest CMR centre), UCL and 10 UK centres, led by Professor James Moon has developed a novel new approach, using Artificial Intelligence, to address this challenge. The team has developed an artificial intelligence solution using deep fully convolutional neural networks (FCNs) with novel task specific architectures, which has been trained on nearly 3000 patients (>2million images) and validated on two datasets: 1) 110 patients scanned twice (for high precision); 2) 1500 patients scanned 5 years ago where survival is known (for prognostication).
This approach beats other methods because of three additional features; 1) geometric transformations to spatially normalise images, improving model specificity and generalisation; 2) an extra neural network for heart long axis function, fusing the geometric transforms for improved precision; 3) modelling output errors and tracing them back to input training, refining human segmentation iteratively until machine vs human performance was indistinguishable, before then testing on the precision/prognostication datasets.
The technology has now been spun-out into a new company, MyCardium AI Ltd, in order to take it to market and to further develop the capabilities for other medical imaging modalities. It is currently raising Series A investment capital in order to accelerate market entry, business scale-up, and further product development. The system is already in use in 17 UK research centres and is being evaluated by global OEMs in cardiac imaging.
For the last two years of its incubation period, Qi3 has been advising the MyCardium team on its strategic development. Qi3’s Robin Higgons undertook market and commercial strategy assessments to help MyCardium evaluate potential market opportunities and align its scientific and development programmes with market needs.
As the market opportunity emerged, Robin gave additional strategy advice on commercialising the MyCardium AI technology and during the spin-out from UCL, Robin advised MyCardium AI’s CEO, Antony Shimmin, on preparations for raising significant investment capital to accelerate MyCardium AI’s development.
“Qi3’s rigorous methodologies for assessing potential markets and Robin’s extensive experience of bringing new technologies to market contributed an incisive and refreshing clarity to MyCardium’s market understanding and to discussions on commercialisation strategies. Robin’s wider experience in advising start-up companies on preparation for raising investment funds was particularly helpful as MyCardium AI looked to its future business models. Robin was a consistent supporter of the team, while also balancing the appropriate level of challenge on business direction and approaches to extract market value and deliver business success. Qi3 always had MyCardium’s objectives in mind and provided clear, insightful advice. I would be pleased to recommend Qi3 to you.