EYE-SCREEN-4-DPN: Development of an innovative Intelligent EYE imaging solution for SCREENing of Diabetic Peripheral Neuropathy

糖尿病是一个主要的全球问题，影响着英国超过490万人。糖尿病每年花费100亿英镑(约占NHS预算的10%)，其中80%是由糖尿病并发症引起的，其中糖尿病周围神经病变(DPN)是最常见的。DPN是由糖尿病引起的神经损伤，可导致麻木、感觉丧失，以及脚、腿和手的神经相关疼痛。由于足部溃疡和最终感染的终末期后遗症，DPN还导致50%-75%的非创伤性肢体截肢。对DPN的筛查将通过在DPN更容易可逆的情况下进行早期干预来改善护理。目前，由于缺乏敏感的、可扩展的以人群为基础的测试，没有有效的DPN筛查方案。首先，没有可靠、易用和准确的诊断工具适合于DPN筛查(从而检测早期DPN)。目前的常规诊断测试是主观的或侵入性的，或无法评估细小的神经纤维(最早受到影响的神经纤维)。我们小组率先使用了一种非侵入性眼科测试，即角膜共焦显微镜(CCM)来成像角膜神经(眼前神经)。CCM是评估早期DPN的一种很好的检查方法。然而，CCM用于DPN筛查一直受到阻碍，因为需要与角膜直接接触，患者不适，视野非常小，检查时间长，需要高水平的操作技能。另一个挑战是缺乏自动化、低成本、可靠和准确的方法来检测DPN并从角膜神经图像中预测DPN的发生。由于主观性的原因，人工评估成本较高，容易出错。我们汇聚了一批在各自领域拥有丰富经验的世界级工程师、科学家和临床医生，开发了第一种为DPN筛查需求量身定做的集成智能成像解决方案。具体目标是1.目的研制一种阶跃式超高分辨率光学相干层析成像(OCT)装置，以取代和克服CCM在非接触成像角膜神经方面的局限性。OCT是一种快速、非侵入性、非接触式成像技术，广泛应用于眼科诊所，包括社区验光师。然而，目前临床上的OCT设备缺乏对角膜神经成像的分辨率。基于我们的专利OCT技术，我们将开发一种新的光学配置，以达到所需的分辨率和速度，以成像大视场的角膜神经，并实现全自动图像采集。开发新的智能算法(软件)来检测和预测护理地点的DPN。分析随着时间推移收集的包括图像在内的大量不同类型的临床数据(纵向数据)的能力仍然是一个挑战。通过利用人工智能的最新进展，我们将生产能够区分患有和不患有DPN的患者、将发展为DPN的患者以及将恶化的患者的工具，从而实现个性化护理和临床管理。3.研制集成OCT设备和AI检测与预测(诊断/预后)的DPN筛查方案原型。这一创新的智能成像解决方案将是可部署的，对临床医生友好。4.为了确认在安特里大学医院开发的创新技术在健康志愿者和糖尿病患者(有无DPN)中的表现，安特里大学医院是DPN和CCM研究的临床卓越中心。总而言之，我们这一雄心勃勃的项目的近期目标是创新的DPN筛查解决方案，而长期目标是可以商业化的完全临床应用的技术。通过我们的创新，及早发现并及时治疗DPN，将预防残疾并挽救生命，为英国的社会和经济带来实质性好处。

Diabetes is a major global problem, affecting over 4.9 million people in the UK. Diabetes costs >£10 billion per annum (~10% of the NHS budget) and 80% of these costs are due to diabetes complications, of which diabetic peripheral neuropathy (DPN) is the commonest. DPN is is nerve damage caused by diabetes and can lead to numbness, loss of sensation, nerve-related pain in the feet, legs, and hands. DPN is also responsible for 50-75% of non-traumatic limb amputations due end stage sequelae of foot ulceration and eventual infection. Screening of DPN will improve care by enabling early intervention where DPN is more readily reversible. At present, there is no effective screening programme for DPN due to a lack of sensitive, scalable population-based tests. First, there are no reliable, easy-to-use and accurate diagnostic tools fit for DPN screening (thus detecting early DPN). Current routine diagnostic tests are subjective or invasive or inability to assess small nerve fibres (the earliest nerve fibres to be affected). Our group has pioneered the use of a non-invasive eye test, namely corneal confocal microscopy (CCM) to image the corneal nerves (nerves at the front of the eye). CCM is an excellent test for the assessment of early DPN. However, the use of CCM for DPN screening has been hindered due to the need for direct contact with the cornea, patient discomfort, very small field of view, prolonged examination time, and requiring a high level of operational skills. The other challenge is the lack of automated, low-cost, reliable and accurate ways detect DPN and predict the occurrence of DPN from corneal nerve images. Manual assessment is expensive and prone to errors due to subjectiveness. We have brought together a group of world-class engineers, scientists, clinicians with extensive experience in their respective fields to develop the first kind of integrated intelligent imaging solution tailored to the needs of DPN screening. The specific objectives are1. To develop a step-change ultrahigh resolution optical coherence tomography (OCT) device to replace and overcome the limitations of CCM for non-contact imaging the corneal nerves. OCT is a fast, non-invasive, non-contact imaging technique that widely used in eye clinics including community optometrists. However, current clinical OCT devices lack the resolution to image the corneal nerves. Based on our patented OCT technology, we will develop a new optical configuration to achieve the desired resolving power and speed for imaging the corneal nerves with a large field of view, and achieve fully automatic image acquisition.2. To develop new intelligent algorithms (software) to detect and predict DPN at the point-of-care. The ability to analyse a large amount of differing types of clinical data collected over time (longitudinal data) including images remains a challenge. By leveraging the recent advances in artificial intelligence, we will produce tools capable of distinguishing between patients with and without DPN, people who will progress to DPN, and in those which it will worsen thus enabling personalised care and clinical management. 3. To produce a prototype DPN screening solution integrating the OCT device and the AI detection and prediction (diagnostic/prognostic). This innovative intelligent imaging solution will be deployable and clinician-friendly. 4. To confirm the performance of the developed innovative technologies in healthy volunteers and people with diabetes (with and without DPN) at the Aintree University Hospital, a centre of clinical excellence in DPN and CCM research.In summary, our immediate goal of this ambitious project is an innovative DPN screening solution, whilst the long-term goal is a fully clinically utilised technology which can be commercialised. Early detection and timely treatment of DPN by our innovations will prevent disability and save lives with substantial benefit to the UK's society and economy.