Image: Adelaide University
An emerging AI tool developed through an Adelaide-led research collaboration could help doctors detect signs of endometriosis from a single pelvic scan in milliseconds, potentially reducing lengthy diagnostic delays and the need for invasive surgery.
Called EndoFusion, the artificial intelligence framework produced results in just 18 milliseconds and correctly distinguished between positive and negative cases 83 per cent of the time during its latest testing.
The tool is being developed by IMAGENDO, an ongoing collaborative study led by Adelaide University researchers, and is designed to identify two common indicators of advanced endometriosis.
Endometriosis is a chronic condition in which tissue similar to the lining of the uterus grows elsewhere in the body. It affects more than 190 million women worldwide and can cause severe abdominal pain, heavy periods, bloating, fatigue, anxiety and infertility.
Despite its prevalence, the condition can be difficult to diagnose. Patients can face years of uncertainty, with confirmation often requiring keyhole surgery to visually identify lesions.
MRI and ultrasound can help detect signs of the condition, but each method has different strengths and can be affected by factors including equipment costs and the experience of the person conducting the scan.
“Current scanning methods each have their own strengths when it comes to detecting two common markers that indicate the likelihood of endometriosis, and patients will often only have access to one of them,” study author Associate Professor Jodie Avery said.
“This means that some patients could be disadvantaged if they are scanned by the less optimal option for their particular signs.”
EndoFusion aims to address this gap by combining information gathered from both MRI and ultrasound imaging.
Researchers developed the framework using four datasets containing more than 9,000 female pelvic MRI scans and more than 800 transvaginal ultrasound sliding scans.
This allows the tool to draw on the strengths of both imaging methods while analysing a single scan, potentially giving clinicians a faster and more comprehensive way to identify signs of endometriosis.
Lead author Dr Yuan Zhang, from Adelaide University’s Robinson Research Institute and the Australian Institute for Machine Learning, said the framework outperformed competing models during the study.
“This is a positive step forward and moves us closer to a future where an AI tool can help clinicians to provide a faster, more accurate diagnosis without the need for surgery,” Dr Zhang said.
While the results are promising, EndoFusion remains in the early stages of development and is not yet being used as a standalone diagnostic tool in clinical practice.
Researchers will next expand the dataset to incorporate additional markers of endometriosis, with the aim of further improving its accuracy.
The team hopes clinicians may eventually be able to use the technology to help determine the likelihood of endometriosis from a single scan. Its multi-imaging approach could also inform future research into other conditions, including gynaecological disorders, prostate and breast cancers and fetal abnormalities.
The study involved researchers from Adelaide University, Flinders University, Benson Radiology, Omni Ultrasound and Gynaecological Care, the University of Surrey, McMaster University Medical Centre and Mohamed bin Zayed University of Artificial Intelligence.
The findings were recently published in Artificial Intelligence in Medicine.
