Researchers from Edith Cowan University (ECU) are creating an artificial intelligence (AI) algorithm that can analyze bone density scans, originally used for detecting spine fractures, to estimate visceral fat levels. This approach promises a quick, painless, and cost-effective method for assessing a dangerous type of fat linked to severe health problems.
Why Visceral Fat Matters
“Ever heard of the sneaky fat that hides deep inside your belly and wraps around your organs? That’s visceral fat,” explained Ph.D. student Arooba Maqsood. “It is strongly linked to serious health problems like heart disease, diabetes, and cancer.”
Maqsood emphasized that obesity remains a major threat to global health, driving morbidity and mortality worldwide. Beyond health, the economic burden is enormous. For Australia alone, obesity cost A$39 billion in 2019, with projections estimating A$228 billion by 2060—3.5% of the country’s GDP. Globally, an estimated 3.7 million deaths each year result from obesity.
Limitations of Current Measurements
Researchers currently estimate visceral fat using body mass index (BMI), waist circumference, or waist-to-hip ratio. However, Maqsood noted that these tools fall short. “They do not distinguish between different types of body fat,” she said. “This oversimplification causes inconsistencies in assessing obesity and its complications, underscoring the need for more precise measurements.”
Although imaging methods like MRI and CT scans can accurately quantify visceral fat, they are expensive. CT scans also expose patients to higher levels of radiation, making them less practical for large-scale screening.
Repurposing DXA Scans for Better Insights
Dual-energy X-ray Absorptiometry (DXA) scans, commonly used to identify spine fractures, provide a unique opportunity. ECU researchers aim to repurpose these lateral spine scans for opportunistic screening of visceral fat. This method could generate valuable health insights without requiring additional tests.
Training AI for Precision
Medicalxpress reports that ECU researchers have already trained their machine-learning algorithm on thousands of DXA scan images. “The next step is to incorporate datasets from around the world so the model learns from the largest, most diverse cohort possible and becomes as effective as possible,” said Dr. Syed Zulqarnain Gilani, senior lecturer and lead AI scientist at ECU.
Presenting at MICCAI 2025
Maqsood will present this groundbreaking research at the International Conference on Medical Image Computing and Computer Assisted Interventions (MICCAI 2025), scheduled to take place in Korea from September 23 to 27.