³ÉÈË´óƬ

Endometriosis is a condition where tissue similar to the lining of the uterus grows outside of the uterus causing severe pain in the pelvis and potentially affecting fertility.

Right now, it takes women in Australia over 6 years to be diagnosed with the condition, even after the onset of symptoms. Researchers at the ³ÉÈË´óƬ’s Robinson Research Institute (RRI) and The Australian Institute for Machine Learning (AIML) are collaborating on an innovative, multidisciplinary project to diagnose the condition much earlier.  

The IMAGENDO® study aims to reduce the diagnostic delay experienced by those with the condition by combining specialist endometriosis transvaginal ultrasound (TVUS) and magnetic resonance images (MRI) using artificial intelligence (AI). With these tools, the IMAGENDO® team can create an algorithm that will determine the probability of a diagnosis of endometriosis much faster. Management plans and treatment can then be undertaken, potentially increasing the patient’s quality of life, improving their fertility chances, and decreasing their levels of pain.

‘If surgery does go ahead, the surgeons have a greater idea of what they will need to do and they can be much more prepared,’ said Dr Jodie Avery, Senior Research Fellow and IMAGENDO® Program Manager. ‘[The doctors] may not have to bring the patient back for further repeat surgeries as these initial surgeries will be undertaken for treatment rather than diagnosis.’

IMAGENDO® was born from a clinical problem observed by Professor Louise Hull, an expert in gynaecology and fertility and Head of the Adelaide Endometriosis Group at RRI. Dr Hull discovered after undertaking a review of 15,000 studies to detect endometriosis before surgery, that transvaginal ultrasound scans and MRI scans outperformed blood and other pathology tests. 

‘I thought that if these tests could be combined digitally, then their detection power would be increased,’ said Dr Hull. ‘So, I sought advice from Professor Gustavo Carneiro, then a professor of medical machine learning at AIML, to see if machine learning could be used to combine the tests.’

‘Gustavo believed it was possible, and after gaining funding in 2021 from the Australian Medical Research Futures Fund, the IMAGENDO® project was born.’

The IMAGENDO team consists of radiologists, gynaecologists, sonologists, computer scientists, sonographers, radiographers, surgeons, and primary health researchers. The team also includes members with lived experience of endometriosis, ensuring that researchers address patient needs effectively.

Yuan Zhang, an AIML PhD student whose PhD work is based on IMAGENDO®, has successfully created the study’s algorithms that look for common signs of endometriosis, including bowel endometriosis. Her algorithms have been created using scans received from people who have had imaging performed and have donated their scans to the project.

‘Our aim is to have one iterative algorithm that will diagnosis endometriosis, using ultrasounds, MRIs, or both,’ said Zhang.

‘There were some challenges using a ‘real world’ dataset that came straight from imaging practices and patients,’ Zhang noted. ‘The data was sometimes mislabelled and there has been frequent misreporting as a result of techniques evolving over time. But I was able to develop methods such as developing an algorithm to extract cleaner labels from the multiple "noisy" labels associated with each training sample to mitigate the ‘noisy’ data.’

Last year, IMAGENDO® won the 2023 ANSTO Eureka Prize for Innovative Use of Technology. This year, the study is about to achieve its first patent and the team is developing its algorithms into a user-friendly platform that can be distributed to imaging clinics and hospitals so that doctors and patients can make use of the technology.

‘In the future, we hope to make an imaging diagnosis of endometriosis so much more accessible for people, especially young people, and those who may live far away from specialist gynaecology services,’ said Dr Avery. ‘We want to empower general practitioners in diagnosing endometriosis and hope to provide an accurate, cheaper, non-invasive alternative to laparoscopic surgery.’

‘Our goal is to eliminate wait times and painful recovery as well as improve the quality of life for people with endometriosis.’

For more information on the IMAGENDO® study, please visit