Digital Phenotyping of Human Placenta Histology

Phenotypic changes to placenta physiology have been linked to many adverse maternal, newborn, and child health outcomes [1,2,3,4,5]. These are associated with serious implications for the future risk of disease and impairments to mother and child [2,6,7]. However, relative to other parts of the body, the placenta is poorly understood [8] and, as such, placenta histopathology requires the manual, labour-intensive effort of a few expert clinicians. There is a lack of computationally efficient, automated tools for the assistance and enhancement of clinical decision-making and pathology assessment.Our research aims to develop an automated pathology phenotyping pipeline linking physiological properties, as detected on human whole slide histology samples, to potential health outcomes. From international collaborators, we have access to richly labelled pathology and phenotype data and the advice of expert clinical pathologists to verify our findings. These collaborators include the Universities of Toronto, University of Tartu, and the Hebrew University of Jerusalem, the Hadassah Medical Organisation, and INTERGROWTH-21st. Such an approach has seen success in other organs for tumour classification [9,10,11], and more recently placenta decidual vasculopathy identification [12], but general automated phenotyping has not been widely investigated. We hope to create a clinically useful tool for assisting in the detection and diagnosis of known placenta pathologies.Aim 1: Extracting Deep Cellular Phenotype:We will develop a self-contained, computationally efficient method for extracting deep cellular phenotype information. This will build upon our existing deep learning pipelines [13] for the localisation and classification of cell types on human and mouse histology samples. It will form the basis for further phenotype investigation and allow for modelling of placental tissue cellular structure.Aim 2: Pathology Phenotyping for Predicting Health Outcomes:We will expand our pipeline with the goal of detecting and classifying pathologies. Using the deep cellular features extracted from Aim 1, we will develop methods, such as modelling cell community structure using graph neural networks, which can detect pathological changes to functional structures in placenta tissue regions. This bottom-up approach may be combined with a top-down segmentation approach for additionally isolating areas of interest. Not only will this allow for the creation of an assistive pathology investigation tool, but it will also provide a new avenue for the exploration of genetic factors and their links to phenotypes.Aim 3: Deep Phenotypes as an Assistive Tool for Pathology Investigation:Working closely with our clinical pathologist collaborators, we will explore how our pipeline can best be used as an assistive tool for pathology investigation. This includes finding the most ethical and clinically useful presentation of model output, e.g. descriptors of phenotypes, pathology predictions with confidence scores, presentation of anomalies, etc., as well as setting realistic computational limits, and A/B testing. This approach will not only ensure that we make the most useful tool, but it will also set out specifications for the creation of other such tools in histopathology analysis.This project falls within the EPSRC Healthcare technologies theme, with a focus on the Optimising Treatment grand challenge and Medical Imaging, Artificial Intelligence, and Image and Computer Vision research areas.Specifically, Aims 1 and 2 contribute towards placenta disease phenotype identification and form a foundational methodology for tools enhancing patient diagnosis. Aim 3 will ensure that the most computationally efficient, ethical, and clinically useful tool is created, leading to cheaper and faster diagnosis thereby increasing the likelihood of successful health outcomes.

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.