Machine learning to classify periportal fibrosis from point-of-care ultrasound to develop clinical decision support systems for schistosomiasis in sub

Schistosomiasis is an infectious disease that is most prevalent in sub-Saharan Africa due to poor access to potable water and sanitation. Over 700 million people are at risk. A parasitic blood fluke causes the conditions related to schistosomiasis. For intestinal schistosomiasis (such as that caused by Schistosoma mansoni), these conditions can include diarrhoea, blood in stool, abdominal distention, gut inflammation, anaemia, enlarged spleens/livers, and in the most severe form periportal fibrosis and portal hypertension. Periportal fibrosis is caused when schistosome eggs are lodged in the portal veins and new branches form to sustain the blood supply to the liver. The mainstay of periportal fibrosis diagnosis is via manual acquisition and grading of liver fibrotic patterns in ultrasound imaging using the World Health Organisation (WHO) Niamey Protocol. This protocol is complex, lengthy, and subject to high inter-reader variability. Advanced expertise in sonography is needed to arrive at the periportal fibrosis grades identified in the Niamey Protocol. In rural poor settings where periportal fibrosis is most prevalent, there is a limited availability of sonographers. The Niamey Protocol was developed for human use before the wide availability of ultrasound devices with the capacity to save images and videos. No updates have been made to the Niamey protocol since its inception in 1996. There is an urgent need to develop automated analyses for periportal fibrosis and investigate whether new features can be detected that are relevant for schistosomal liver disease.

To solve this problem, an automated analysis pipeline will be developed that will take an ultrasound video generated from a specific set of simple-to-acquire sweeps of the liver. The main goals will be to produce assistive aids for identifying where fibrosis occurs, produce an automated version of the grading of liver fibrosis according to the Niamey Protocol, and to compare new features/classifications detected to the existing grading procedure. The first step will be to identify the video frames with the best representation of different anatomies in the video, using automated segmentation with supervised and unsupervised approaches. Next, a classifier will be developed to look for features in these images that indicate different grades of fibrosis. The dataset to be analysed will come from SchistoTrack, which is a human participant cohort that was set up to investigate liver fibrosis progression in rural populations in Western and Eastern Uganda.

Aims

1. To develop a pipeline for automated liver segmentation, identifying relevant anatomy for periportal fibrosis.
2. To identify features of fibrosis grades from ultrasound videos and images focused on liver anatomies with validation from a trained sonographer to promote model interpretability and trustworthiness.
3. To test the performance of the automated analysis pipeline by identifying robust evaluation metrics apart from accuracy.
4. To analyse data from different sites/groupings and consider how these fit into clinical practice in Uganda, including a pilot test in the study areas.

This research will make progress in automated abdominal ultrasound segmentation and fibrotic feature detection in the livers. These machine learning-based topics are poorly understood. The main goal is for the analysis and assistive technology pipelines to be accurate, flexible and interpretable, meaning they could be used to aid and train sonographers in sub-Saharan Africa and to contribute to addressing the clinical need for quicker diagnosis and intervention.

This project falls within the EPSRC's healthcare technologies theme. It is a collaboration with the Uganda Ministry of Health through the Oxford-Uganda Collaboration on Schistosomiasis and the SchistoTrack human participant cohort.

In the same way that bioinformatics has transformed genomic research and clinical practice, health data science will have a dramatic and lasting impact upon the broader fields of medical research, population health, and healthcare delivery. The beneficiaries of the proposed training programme, and of the research that it delivers and enables, will include academia, industry, healthcare, and the broader UK economy.

Academia: Graduates of the training programme will be well placed to start their post-doctoral careers in leading academic institutions, engaging in high-impact multi-disciplinary research, helping to build training and research capacity, sharing their experience within the wider academic community.

Industry: Partner organisations will benefit from close collaboration with leading researchers, from the joint exploration of research priorities, and from the commercialisation of arising intellectual property. Other organisations will benefit from the availability of highly-qualified graduates with skills in big health data analytics.

Healthcare: Healthcare organisations and patients will benefit from the results of enabled and accelerated health research, leading to new treatments and technologies, and an improved ability to identify and evaluate potential improvements in practice through the analysis of real-world health data.

Economy: The life sciences sector is a key component of the UK economy. The programme will provide partner companies with direct access to leading-edge research. Graduates of the programme will be well-qualified to contribute to economic growth - supporting health research and the development of new products and services - and will be able to inform policy and decision making at organisational, regional, and national levels.