When Alejandra Coronel-Zegarra peers into a micro-CT scan of a coral colony, she sees what the naked eye could never catch: invisible fractures in the skeleton, shifts in density, and tiny pores widening like cracks in aging bone. Now, thanks to a breakthrough combining 3D imaging and artificial intelligence, researchers at Florida Atlantic University in Boca Raton are reading those hidden signs with startling precision—and catching coral diseases before they deal fatal blows.
Since 2014, Stony Coral Tissue Loss Disease has been sweeping across Florida's reefs and spreading through the Caribbean, leaving behind ghostly white skeletons where thriving colonies once teemed with life. What made the crisis especially frustrating was how little scientists understood about what was happening at the microscopic level. Traditional methods of studying coral skeletons were slow, invasive, and often missed the subtle structural changes that preceded visible die-offs.
FAU's team turned to microcomputed tomography—the same X-ray technology used in hospital CT scans, but far more powerful. The high mineral content of coral skeletons makes them ideal for this imaging technique, which can reconstruct a colony's entire 3D architecture without cutting into it. Working from the FAU High School Owls Imaging Lab, researchers scanned both healthy specimens and those affected by SCTLD, focusing on two key species: Montastraea cavernosa and Porastraea.
To make sense of the massive datasets generated by these scans—thousands of cross-sectional images per colony—the team trained deep learning models to automatically distinguish coral skeleton from empty pore space. They tested three convolutional neural network architectures, ultimately finding that Attention U-Net delivered the best results: more than 98% accuracy in identifying structural features, completing full segmentation in just seven hours compared to seventeen hours for a standard U-Net model.
"Without high-resolution, 3D insights, scientists cannot fully understand how disease, warming oceans and other stressors compromise reef survival," said Vivian Merk, an assistant professor in FAU's Department of Chemistry and Biochemistry and corresponding author of the study published in the Journal of Structural Biology. "By uncovering these hidden changes in porosity, density and skeletal thickness, we can see exactly how diseases like Stony Coral Tissue Loss Disease alter the physical integrity of corals."
The analysis revealed clear structural differences between healthy colonies and those in early disease stages—changes invisible to conventional monitoring but unmistakably captured by the AI. For conservation teams racing to save vulnerable reef sites, this means the difference between reacting to visible damage and catching problems while intervention is still possible. Coronel-Zegarra, a PhD candidate who won the 2025 Microscopy and Microanalysis Student Award for this work, sees the technology as a turning point: "By combining micro-CT with deep learning, we can automatically detect subtle changes in the skeleton caused by disease—details that are nearly impossible to see manually."
The team is now working to expand the approach across more coral species and larger datasets, bringing automated, non-invasive health monitoring closer to routine use in reef conservation. It's a small window into a coral's skeleton—but it might just be enough to save the reef.