DEFEND
Research type
Research Study
Full title
Developing an US-MRI biomarker fusion model for Endometriosis
IRAS ID
281860
Contact name
Ippokratis Sarris
Contact email
Sponsor organisation
Perspectum Ltd
Duration of Study in the UK
2 years, 11 months, 31 days
Research summary
Research Summary:
This study will be an observational study conducted at Kings Fertility involving 100 women with endometriosis. It is primarily designed to collect information from women with endometriosis and use data from ultrasound (US) scans and magnetic resonance imaging (MRI) to develop a database representing the variability of the disease.
Currently the first recommended diagnostic investigation for endometriosis is a US scan or a MRI followed by a diagnostic surgery - laparoscopy (the insertion of a tube into the abdomen under general anaesthetic). Unfortunately, the current mode of practice is subjective and its utility at predicting surgical findings of endometriosis is limited. Furthermore, accurate prediction using US and MRI is limited to selected tertiary units where high-level expertise exists. The latest report from the National Institute of Clinical Excellence (NICE) reports a time delay of around 7.5 years before a confirmed diagnosis of endometriosis.
A model that could accurately predict surgical findings of endometriosis would be of significant clinical and economical benefit. With the information collected, we propose to develop an intelligent predictive model that will accurately predict endometriosis by fusing different MRI sequences; as well as MRI and ultrasound image fusion and ultrasound/ultrasound fusion. In addition to image fusion, we aim to incorporate a biomarker (MDNA) as an additional predictive tool. Its significance, when used in combination with imaging findings to predict endometriotic nodule proliferation, has not been robustly studied before.
By validating our predictive model with surgical findings, we aim to develop a widely available diagnostic tool based on image fusion and biomarker using computer modelling. This will increase confidence and access to advanced imaging for non-experts, allow clinicians to accurately predict surgical findings as well as reduce time to diagnosis.
Summary of results :
This project formed part of the National Consortium of Intelligent Medical Imaging (NCIMI), which was a UK Government initiative that aimed to accelerate the uptake of AI-based medical imaging in the NHS. This project is a collaboration between King's Fertility, King's College Hospital NHS Foundation Turst, Perspectum, and GE Healthcare.
This project looked at endometriosis, a condition where tissue from the uterus is found in other parts of the body. It causes chronic pelvic pain and often difficulty with becoming pregnant.Several imaging techniques, including ultrasound, computerised tomography (CT) and magnetic resonance imaging (MRI), have been previously tried for detecting endometriosis.
Acording to NICE guidelines, endometriosis is diagnosised by a surgical inspection of the pelvis. It is hoped that less invasive methods may be used for diagnosis, including ultrasound and MRI scanning.
In this study, we developed a database of ultrasound and MRI data of endometriosis showing researchers the variability of disease and imaging seen in gynaecological practice, as a way to speed up building imaging tools for diagnosing endometriosis. We imaged 64 patients that attended King's College Hospital and the King's Fertility clinic prior to their planned surgery for endometriosis, and collected pre-operative imaging data, clinical histories, and subsequent surgical findings. We asked the patients about how the dealt with the pain, and how difficult it has been to become pregnant.
This data has been stored initially in NCIMI and then later transferred to BMRC in Big Data Institute.
The data has continued to be used for research purposes, making progress in identifying different types of endometriosis, specifically endometriomas.This study has built an imaging database that contains information about patients with known endometriosis and their clinical history. It is hoped that research uses this data to further develop non-invasive methods of identifying endometriosis so that women do not need to have surgery in order to start treatment for this condition.
REC name
West of Scotland REC 5
REC reference
20/WS/0111
Date of REC Opinion
27 Aug 2020
REC opinion
Further Information Favourable Opinion