CoralLite: μCT Reconstruction of Coral Colonies from Individual Corallites

2026-05-14

AI Summary

This study addresses the challenge of reconstructing and tracking individual corallite structures from micro-computed tomography (μCT) scans to elucidate the relationship between polyp budding history and skeletal growth. To this end, we introduce CoralLite, the first dataset comprising 697 slices with 37 meticulously annotated samples, and propose a novel 3D reconstruction method based on a hybrid V-Trans-UNet architecture that integrates volumetric segmentation with cross-slice linking. This approach enables, for the first time, fully automated 3D corallite modeling solely from μCT scans. Leveraging weakly supervised pretraining and topology-aware fine-tuning, our method achieves a topological accuracy of 0.94 and an average Dice score of 0.77 on in-sample data, with a cross-sample Dice score of 0.63. The code, models, and dataset are publicly released.

Abstract

The life history of an individual coral is archived within the accreting skeleton of the colony. While reef-forming coral colonies (e.g. massive \emph{Porites} sp.) may live for hundreds of years and deposit calcareous structures many metres in height and width, their living tissue is a thin outer surface layer comprised of asexually-dividing polyps that only survive a few years. To understand the rate and timing of polyp division and the consequences for colony skeletal growth, scientists need to track the skeletal corallite deposited around each polyp. Here we propose CoralLite, an annotated μCT scan dataset of entire calcareous skeletons and an associated, first corallite deep learning reconstruction baseline. CoralLite combines fully quantified volumetric segmentations with cross-slice linking for visualisations of 3D models for each corallite up to colony scale. For segmentation, we propose and evaluate in detail a hybrid V-Trans-UNet architecture applicable to segmenting tiled μCT virtual slabs of \emph{Porites} sp. colonies. The model is pre-trained on weakly annotated data and topology-aware fine-tuned using fully annotated slice sections with 8k+ manual corallite region annotations. On unseen slices of the same colony, the resulting model reaches 0.94 topological accuracy at mean Dice scores of 0.77 on the same colony and projection axis, and 0.63 mean Dice scores on a different, biologically unrelated specimen. Whilst our experiments are limited in scale and context, our results show for the first time that visual machine learning can effectively support full 3D individual corallite modelling from μCT scans of coral skeletons alone. For reproducibility and as a baseline for future research we publish our full dataset of 697 μCT slices, 37 partial or full slice annotations, and all network weights and source code with this paper.

Problem

• corallite
• coral skeleton
• μCT reconstruction
• 3D modeling
• polyp division

Innovation

• corallite segmentation
• μCT reconstruction
• V-Trans-UNet
• 3D coral modeling
• weakly supervised learning