CALOPUS - Computer Assisted LOw-cost Point-of-care UltraSound

点击翻译按钮获取中文摘要

The CALOPUS project is an inter-disciplinary UK-India collaboration that aims to automate the analysis of antenatal ultrasound images to facilitate referral to the appropriate level of pregnancy care in rural settings, an important unmet need in many LMICs. Unlike in the western world, many women in India do not have an antenatal ultrasound check-up scan in the first or second trimester. Without such a check-up, health problems of the fetus and mother can go unnoticed, leading, in the worst case, to a still-birth or preterm birth. India has the highest number of still births and preterm births in the world (Lancet 2O15, 2012). Thus the ultimate clinical aim motivating this work is to reduce the burden of adverse pregnancy events and improve the health of future generations. The approach we are taking is to use a low-cost portable imaging technology, ultrasound imaging, as the basis for a novel point-of-care (POC) system that enables a healthcare worker, with limited or no knowledge of ultrasound, to refer a patient to a clinical unit or hospital. Ultrasound scanning normally requires skill to acquire images which is a barrier to use. Our idea is to combine capture of ultrasound videos with machine learning (deep learning) to automatically interpret the content of the ultrasound videos and make the recommendation to the non-expert user. This project explores how we can do this, and will look, as well, at how we can make the technological solution culturally acceptable, and usable in the intended clinical environment. In addition to the primary scientific endeavour, an important secondary objective will be capacity-building in point-of-care imaging in the India partner team and counter-mentoring of UK investigators in health (particularly in factors related to healthcare technology needs and adoption), and research and cultural issues that are at play in LMICs. This cross-dialogue will strengthen the international team and build mutual respect and complementary expertise.