Artificial Intelligence assisted echocardiography to facilitate accurate image capture and transmission for congenital heart defects diagnosis in Sub-Saharan Africa

This project plans to develop an AI driven model that will facilitate echocardiography scan of neonates (0-28 days) by non-experts, by allowing automatic image/video clip capture. The proposed solution will essentially allow for a freehand scan, removing the need for the sonographer to be skilled in the subtle and complex movements of the probe that are necessary for the acquisition of standard cardiac image/video clips. The performance of the model will continuously improve as more data is acquired during use in the field. This means local low skilled sonographers who currently are only able to perform routine pregnancy ultrasound scans, will now be able conduct postnatal echocardiography scans for neonates suspected of having a CHD after pulse oximetry screening, allowing them to capture accurate images/video clips that can be transmitted to a remote expert for diagnosis confirmation. This breakthrough will enhance neonatal CHD diagnosis and care in SSA in two major ways: 1.Eliminate the extra burden on false positive CHD cases from pulse oximetry screening who have to travel thousands of kilometres to expert facilities to know their status. 2.Allow for a risk-based stratification treatment process where non-critical cases can be treated locally as directed by the remote specialist and only critical cases requiring expert care are referred to the expert centres. This reduces the cost and burden/risk of travelling for non-critical cases, allows for more attention to and early commencement of treatment for critical cases, and reduces workload on experts. In summary, our proposed technology will help streamline the CHD diagnosis process by maximizing the use of local available low skilled sonographers, reduce cost and burden/risk of unnecessary travel for the particularly vulnerable and fragile baby, and reduce the workload and optimize the performance of the limited available experts, while increasing their overall efficiency and accuracy of diagnosis.