Automated Image-Diagnosis for Female Genital Schistosomiasis

点击翻译按钮获取中文摘要

Context and Potential Impact Schistosomiasis is the second most devastating tropical disease in terms of public health burden and economic growth. More than 240 million people worldwide suffer from schistosomiasis. Merck has been fighting against this Neglected Tropical Disease with its partner, the World Health Organization, since 2007 in the scope of its Merck Schistosomiasis Elimination Program and donates up to 250 million praziquantel tablets for treatment per year. Merck and WHO committed themselves to continue their work in fighting the disease until its elimination. Female Genital Schistosomiasis (FGS) is acquired from untreated infection with Schistosoma haematobium, which results from contact with contaminated freshwater bodies used by rural communities, specifically by women and girls, for daily chores and activities. FGS affects 56 million African women and girls. FGS can gradually evolve towards reproductive organ damage characterized by sub-fertility or infertility, ectopic pregnancy, spontaneous abortion, premature birth, low birth weight and maternal death. Additionally, there is concern and biological plausibility that FGS increases susceptibility to HIV and HPV infection, and cervical cancer. Nevertheless, FGS remains largely underreported, under- and misdiagnosed and untreated. FGS is characterized by a pattern of lesions that manifest at the cervix and the vagina, such as homogeneous and grainy sandy patches. At present, a key challenge in eliminating FGS is that there is no gold-standard diagnostic suitable for endemic settings. Laboratory diagnostics are currently inadequate and a low index of suspicion among healthcare professionals can lead to misdiagnosis of sexually transmitted infections. An automated image diagnostic system could improve FGS treatment and research programs by providing a low-cost, point-of-care diagnostic aid for local healthcare professionals. This would help diagnose FGS, leading to better patient outcomes, and help awareness and treatment programs to better target at-risk communities. Aims and Objectives The doctoral research project aims to build an open-source, automated, image-based, digital point-of-care diagnostic system for FGS to operate in resource-poor settings. This tool could subsequently be trialed in a global health setting with Merck's clinical partners. Objectives will include designing and implementing a lightweight network architecture for use in an ultra-low powered environment, evaluating data augmentation methods for enriching limited clinical datasets and investigating best practice for model explainability and diagnostic reporting to build trust with clinicians and end-users. Through this research project and the creation of a well-documented, open-source tool, a secondary goal will be to enable reproducibility of similar solutions in other resource-constrained global health applications. EPSRC Alignment and Industrial Collaboration This project falls within the EPSRC Medical Imaging research area, part of the broader Healthcare Technologies portfolio, and targets the Earlier Diagnosis research priority. The project is in collaboration with Merck.