Coral Damage Detection: FAU Study Reveals Hidden Damage in Stony Corals Using 3D Imaging and AI

🪸 The Urgent Crisis Gripping Florida's Coral Reefs

Florida's coral reefs, stretching over 360 miles along the southeastern coast, form one of the world's most biodiverse marine ecosystems. These underwater structures support over 1,700 fish species, provide coastal protection against storms valued at billions annually, and sustain tourism economies exceeding $8 billion yearly. However, since 2014, Stony Coral Tissue Loss Disease (SCTLD), a mysterious and lethal pathogen, has ravaged these reefs, affecting more than 20 of Florida's approximately 45 reef-building coral species and impacting up to 95% of some reef areas.6994

SCTLD manifests as white, bleached tissue patches that spread rapidly, often killing entire colonies within weeks. Species like the great star coral (Montastraea cavernosa) and mustard hill coral (Porites astreoides) have been hit hard, leaving behind dead skeletons that disrupt habitats for fish, lobsters, and other marine life. By 2026, the disease has spread to 22 Caribbean countries, underscoring its pandemic-like threat to global reef health. Traditional monitoring relied on surface observations, missing the subtle internal skeletal damage that compromises long-term reef resilience.

Spotlight on Stony Coral Tissue Loss Disease (SCTLD)

Stony Coral Tissue Loss Disease, first detected off Miami in 2014, is characterized by rapid tissue necrosis, exposing the white skeleton underneath. Unlike bleaching from heat stress, SCTLD persists year-round and transmits via water currents, potentially involving bacterial pathogens though the exact cause remains elusive. In Florida's Coral Reef Tract, prevalence has reached 55% in some sites, lasting months before declining, but with lasting ecosystem scars.70

Quantitative impacts include colony mortality rates up to 66% for vulnerable species within a year of infection. Reef-building stony corals, or scleractinians, deposit calcium carbonate skeletons that form the reef framework. Disease weakens this by altering microstructure— increasing porosity, reducing density—making colonies brittle and prone to breakage during storms. Prior studies used 2D histology or low-resolution scans, but lacked the precision to quantify these changes comprehensively.

• Transmission: Likely waterborne, affecting both shallow and deep reefs.
• Susceptibility: Massive corals like M. cavernosa show variable resistance; columnar species succumb faster.
• Ecological fallout: Loss of 30+ species locally, shifting to algae-dominated reefs.

FAU's Pioneering Approach: Micro-CT Meets Artificial Intelligence

Researchers at Florida Atlantic University (FAU) have developed a game-changing method using X-ray microcomputed tomography (micro-CT) paired with deep learning artificial intelligence (AI). Published April 14, 2026, in the Journal of Structural Biology, the study titled "Leveraging deep learning semantic segmentation for imaging coral skeletons" reveals hidden skeletal damage invisible to the naked eye.Read the full paper here95

Micro-CT, housed in FAU's High School Owls Imaging Lab, fires X-rays through samples to create high-resolution 3D models down to micrometers, ideal for corals' dense calcium carbonate. AI automates analysis, segmenting skeleton from pores with unprecedented speed and accuracy.

Breaking Down the Methodology: From Scan to Insight

The process begins with sample preparation: small cores from healthy and SCTLD-affected M. cavernosa and healthy P. astreoides are scanned non-destructively. Micro-CT generates thousands of 2D slices, reconstructed into 3D volumes.

Deep learning enters via convolutional neural networks (CNNs)—specifically U-Net variants: standard U-Net, U-Net++, and Attention U-Net. Trained on annotated datasets, these models learn to classify voxels as skeleton or pore by recognizing textural patterns. Attention U-Net excelled, achieving over 98% segmentation accuracy in just 7 hours per dataset, versus 15-17 hours for others.

1. Imaging: Micro-CT scan (high contrast due to mineralization).
2. Preprocessing: Noise reduction, normalization.
3. Training: Supervised on labeled slices from multiple corals.
4. Segmentation: Pixel-level classification.
5. Quantification: Porosity (pore volume fraction), bulk density, thickness maps.

This step-by-step pipeline quantifies changes: diseased corals showed increased porosity (up to 20% higher in some regions), decreased density, and thinner septa—the radial walls defining corallites.94

Key Findings: Unveiling Microscopic Devastation

The FAU study exposed profound skeletal alterations. Healthy M. cavernosa exhibited uniform, compact pores; SCTLD-affected ones displayed irregular, enlarged pores reducing structural integrity by 15-25%. Bulk density dropped significantly, correlating with observed fragility in field studies. Species comparisons revealed P. astreoides' denser skeletons confer partial resistance.

• Porosity: +18% average in diseased vs. healthy.
• Density: -12% in affected skeletons.
• Thickness: Septal thinning by 10-30 micrometers.
• Accuracy: 98.5% across models, validated statistically.

"Micro-CT gives us a window into the coral skeleton in a way that’s never been possible before," noted lead author Alejandra Coronel-Zegarra. These metrics link disease to mechanical weakness, explaining higher breakage rates post-infection.

Meet the FAU Innovators Driving This Research

Alejandra Coronel-Zegarra, Ph.D. candidate in FAU's Department of Chemistry and Biochemistry, led the effort, earning the 2025 Microscopy and Microanalysis Student Award. Corresponding author Vivian Merk, Ph.D., assistant professor spanning Chemistry, Biochemistry, Ocean, and Mechanical Engineering, emphasizes interdisciplinary impact. Jamie Knaub, Biology Ph.D. candidate and lab assistant, handled imaging, while Abhijit Pandya, professor in Electrical Engineering, Computer Science, and Biomedical Engineering, optimized AI models.

Funded by NSF, FAU's College of Engineering seed grants, and I-SENSE Institute, this collaboration exemplifies FAU's strength in integrative science.FAU News Release

Transformative Implications for Reef Monitoring and Conservation

This technology shifts coral health assessment from surface-level to structural deep dives, enabling early detection of at-risk reefs. Managers can prioritize interventions like antibiotic pastes or resistant genotype outplanting. For Florida, protecting the $8.5 billion reef economy demands such precision tools. Globally, it informs responses to parallel threats like ocean acidification eroding skeletons similarly.

"Our analyses provide a clearer, quantitative picture of how environmental stressors reshape coral skeletons," said Merk. By mapping vulnerabilities, it supports resilient hybrid breeding programs at FAU Harbor Branch.

FAU's Broader Contributions to Marine Research

FAU Harbor Branch Oceanographic Institute leads SCTLD response, developing 3D photogrammetry for disease tracking since 2021. Recent grants fund mesophotic reef studies, revealing deeper refugia. In higher education, FAU trains students in AI-marine applications, fostering careers in oceanography and engineering. Programs like Vertically Integrated Projects engage undergrads in real-world reef science.

Future Horizons: Scaling AI for Global Reefs

Next steps include portable micro-CT for field use, multi-species expansion, and integration with drones for reef-scale mapping. Collaborations with NOAA aim to model SCTLD spread using skeletal data. Challenges like scan resolution limits and AI training data scarcity persist, but solutions like transfer learning promise scalability. By 2030, such tools could halve monitoring costs, boosting restoration success rates from 20% to over 50%.

Photo by Francisco Jesús Navarro Hernández on Unsplash

Stakeholder Perspectives and Actionable Insights

Conservationists hail the method's non-invasiveness; policymakers eye it for Reef Advisory Committee strategies. Divers and ecotour operators stress economic stakes—lost reefs cost 10,000 jobs. Actionable: Fund AI training datasets, deploy at key sites like Dry Tortugas, integrate into National Coral Reef Management Plans. For academics, it opens doors to study acidification, pollutants' synergistic effects.79