Development of smartphone imaging technology for screening maternal and neonatal anaemia

研究背景的简要描述，包括潜在影响贫血影响世界人口的四分之一，孕妇的患病率更高，为42%。该项目的目的是使用负担得起的智能手机成像作为孕产妇和新生儿贫血的诊断策略。目前，诊断需要血液检查，这是昂贵的和侵入性的。在缺乏有效运输和医院系统的国家和地区，例如经历政治动荡的国家，可能无法进行血液检测。及时诊断贫血可以使治疗更容易（例如，通过补充铁），更有可能有效。该方法是基于眼睛和眼周区域的成像，包括下眼睑。我们的目标是开发一款名为“孕产妇和新生儿综合护理”的智能手机应用程序，该应用程序将提供有关贫血和黄疸的诊断信息（在这个项目之外），用于怀孕期间的妇女、出生后不久的婴儿和学龄前儿童。该小组以前曾利用类似的技术检测新生儿黄疸，这为临床研究建立了联系，并获得了广泛的媒体关注。目的和目标-主要目标是开发用于筛查孕产妇贫血的孕产妇和新生儿综合护理（iMANC）应用程序，供训练有素的卫生保健工作者、家长和患者使用。-具体目标是：o改善图像采集和预处理：-通过机器学习开发巩膜分割，以识别图像的适当区域进行进一步分析-开发质量保证的自动化流程，以检测和替换不适合分析的图像-调查和利用经济实惠的智能手机成像技术（如双摄像头传感器）的进步，以提高图像质量-改善图像分析：-使用通过对临床研究的设计、实施和分析，对iMANC应用的准确性进行统计分析。了解该技术在临床应用中面临的挑战，包括伦理和财务方面的问题。以及对临床医生工作流程的影响，这可能对更广泛的使用构成障碍。研究方法的新颖性尽管研究小组已经使用这种技术进行了初步研究，但初步研究并未在所有亚人群中起作用。为了克服这一问题，研究将从两个方面开发新的技术：(1)人机交互，用于指导图像采集，自动获取收集数据中的感兴趣区域，并以可访问和负责任的方式呈现输出数据；(2)优化诊断算法。这包括开发控制环境光效果的新技术，以及使用尚未广泛使用的新智能手机相机技术。与EPSRC的战略和研究领域保持一致该项目属于医疗保健技术主题，特别是与临床技术、医学成像和分析科学领域保持一致。这是因为它旨在开发新的技术，通过控制医学图像的获取来分析临床环境中的生物系统。通过开发自动图像分割算法，该项目也可能与人工智能技术主题保持一致。合作者：Sara Hillman博士（UCLH）， Judith Meek博士（UCLH）， Vatsla Dadhwal博士（AIIMS）， Anubhuti Rana博士（AIIMS）公司：全印度医学科学研究所，加纳大学，大阿克拉地区医院

Brief description of the context of the research including potential impactAnaemia affects a quarter of the world population with a greater prevalence of 42% in pregnant women. The aim of the project is to use affordable smartphone imaging as a diagnostic strategy in maternal and neonatal anaemia. At present, diagnosis requires a blood test, which is costly and invasive. Blood tests may not be accessible in countries and regions without effective transport and hospital systems, such as in countries experiencing political unrest. A timely diagnosis of anaemia can allow for easier treatment (for example, via iron supplementation) with a greater chance of efficacy.The approach is based upon imaging of the eye and the periocular region, including the lower eyelids. We aim to develop a smartphone application called the integrated maternal and neonatal care app which will provide diagnostic information on anaemia and jaundice (outside this project), to be used on women during pregnancy, their babies shortly after birth, and preschool-aged children. The group have previously utilised similar technology in detection of neonatal jaundice, which has established links for clinical studies and achieved widespread media attention.Aims and Objectives- The primary objective is to develop the integrated maternal and neonatal care (iMANC) app for screening for maternal anaemia, which can be used by trained healthcare workers, parents and patients.- The specific objectives are to:o Improve image acquisition and pre-processing:- Develop sclera segmentation by machine learning in order to identify appropriate regions of the image for further analysis- Develop automated processes for quality assurance to detect and replace images which are not suitable for analysis- Investigate and utilise advances in affordable smartphone imaging technology (such as dual camera sensors) in order to improve image qualityo Improve image analysis:- Use the programmable camera settings to investigate techniques for minimising the effects of varying ambient lighting- Investigate additional parameters within the acquired image which might improve diagnostic abilityo Clinical validation:- Statistically analyse the accuracy of the iMANC application through design, implementation, and analysis of clinical studies- Understand the challenges to clinical use of this technique, including both ethical and financial concerns, as well as effects on clinician workflow which may pose barriers to wider useNovelty of Research MethodologyAlthough the research group have used this technology to carry out pilot studies, the pilot studies have not worked in all subpopulations. In order to overcome this, the research will develop novel techniques with two areas: (1) human-computer interaction, for guiding image acquisition, automatically obtaining regions of interest in the gathered data, and presenting the output data in an accessible and responsible manner; and (2) optimising the diagnostic algorithm. This includes development of new techniques for controlling for the effect of ambient light and using new smartphone camera technologies which have not been used widely.Alignment to EPSRC's strategies and research areasThis project lies within the healthcare technology theme, in particular aligning with the areas of clinical technology, medical imaging, and analytical science. This is because it aims to develop novel techniques to analyse a biological system in a clinical setting through controlled acquisition of a medical image. This project may also align against the artificial intelligence technologies theme through development of automatic image segmentation algorithms.Any companies or collaborators involvedCollaborators: Dr Sara Hillman (UCLH), Dr Judith Meek (UCLH), Dr Vatsla Dadhwal (AIIMS), Dr Anubhuti Rana (AIIMS)Companies: All India Institute of Medicine Science, University of Ghana, Greater Accra Regional Hosp