Exploring Nile Red and machine learning for microplastics detection in Tridacna maxima
5 affiliations
- 1Université de Polynésie Française, UMR SECOPOL, Tahiti, Polynésie Française
- 2AI Freelancer Papeete, Tahiti, Polynésie Française
- 3Aix-Marseille Université, LCE, Aix-en-Provence, France
- 4Université de Bordeaux, CNRS, UMR EPOC 5805, Pessac, France
- 5Institut de Recherche pour le Développement, UMR SECOPOL 241, Tahiti, Polynésie Française
Abstract
Small Island Developing States (SIDS) face unique challenges for microplastics (MPs) monitoring due to limited infrastructure and resources. In this context, we propose and test innovative approaches toward a standardized, low-cost methodology for quantifying MPs in SIDS. We evaluate the giant clam T. maxima as a bio-integrator, combining Nile red (NR) fluorescence staining with automated machine-learning detection. We optimized a digestion protocol using KOH and HNO3 for T. maxima viscera, and developed a DAPI-guided multi-spectra composite imaging approach based on triband fluorescence (DAPI, FITC, TRITC), to enhance polymer detection while reducing blooming artifacts. A semi-automated annotation pipeline using Labkit interactive segmentation with CLIP/UMAP clustering efficiently generated training data from 6711 fluorescence images. A U-Net model was trained on composite images to segment fluorescent particles. The workflow was applied to giant clams from three French Polynesian islands (Makemo, Hao, Tubuai), and NR-based estimates were validated against µFTIR spectroscopy. The model achieved F1-scores of 0.741 for giant clam samples and 0.657 for controls, comparable to human annotation (F1 = 0.680). MPs were detected across all islands, with highest concentrations in gills (16.9–52.7 particles·g−1 wet weight) compared to viscera (2.5–11.0 particles·g−1 ww). µFTIR validation revealed that NR overestimates MP counts (µFTIR: 0.80 ± 0.16 particles·g−1 ww at Tubuai), primarily due to false positives from proteins, cellulose, and stearates. In Tubuai, polyamide (28.9%), PVC (12.6%), and polystyrene (10.7%) were the dominant polymers, suggesting contributions from fishing gear, agriculture, and household waste. While NR-based quantification overestimates absolute MP counts, the automated pipeline demonstrates potential for high-throughput image processing, reproducible sample analysis, and methodological standardization. This workflow represents a first step toward scalable, low-cost approaches for MPs monitoring in insular systems, highlighting areas for further calibration and optimization. Future work should refine fluorescence thresholds and expand validation across species and locations.
Related
- Exploring lithium diffusion in LiF with machine learning potentials: from point defects to collective ring diffusion
- Machine Learning-Driven Multiplexed Biomarker Detection with Polymer-Enhanced Electrochemical Sensors
- Metabolism-Based Bacterial Detection in Water Using First-Principles Modeling and Predictive Machine Learning
- Multimodal Representation Learning for Exploring Natural Product Chemical Space
- Exploring the photocatalytic activity and acidity of nanocatalyst based on green copper oxide nanoparticles and nanozeolites: Exploratory Data Analysis, Computer Vision and Deep Learning
BibTeX
@article{godere2026exploring,
title = {{Exploring Nile Red and machine learning for microplastics detection in Tridacna maxima}},
author = {Irène Godéré and Taiamiti Edmunds and Nabila Gaertner-Mazouni and Fiona Gimenez and Pascal Wong-Wah-Chung and Stéphanie Lebarillier and Magalie Baudrimont and Chloé Pupier and Nicolas Maihota and Jean-Claude Gaertner},
journal = {PLOS One},
year = {2026},
month = sep,
doi = {10.1371/journal.pone.0357014},
url = {https://doi.org/10.1371/journal.pone.0357014},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Exploring Nile Red and machine learning for microplastics detection in Tridacna maxima
AU - Irène Godéré
AU - Taiamiti Edmunds
AU - Nabila Gaertner-Mazouni
AU - Fiona Gimenez
AU - Pascal Wong-Wah-Chung
AU - Stéphanie Lebarillier
AU - Magalie Baudrimont
AU - Chloé Pupier
AU - Nicolas Maihota
AU - Jean-Claude Gaertner
PY - 2026
DA - 2026/09/16
JO - PLOS One
DO - 10.1371/journal.pone.0357014
UR - https://doi.org/10.1371/journal.pone.0357014
AB - Small Island Developing States (SIDS) face unique challenges for microplastics (MPs) monitoring due to limited infrastructure and resources. In this context, we propose and test innovative approaches toward a standardized, low-cost methodology for quantifying MPs in SIDS. We evaluate the giant clam T. maxima as a bio-integrator, combining Nile red (NR) fluorescence staining with automated machine-learning detection. We optimized a digestion protocol using KOH and HNO3 for T. maxima viscera, and developed a DAPI-guided multi-spectra composite imaging approach based on triband fluorescence (DAPI, FITC, TRITC), to enhance polymer detection while reducing blooming artifacts. A semi-automated annotation pipeline using Labkit interactive segmentation with CLIP/UMAP clustering efficiently generated training data from 6711 fluorescence images. A U-Net model was trained on composite images to segment fluorescent particles. The workflow was applied to giant clams from three French Polynesian islands (Makemo, Hao, Tubuai), and NR-based estimates were validated against µFTIR spectroscopy. The model achieved F1-scores of 0.741 for giant clam samples and 0.657 for controls, comparable to human annotation (F1 = 0.680). MPs were detected across all islands, with highest concentrations in gills (16.9–52.7 particles·g−1 wet weight) compared to viscera (2.5–11.0 particles·g−1 ww). µFTIR validation revealed that NR overestimates MP counts (µFTIR: 0.80 ± 0.16 particles·g−1 ww at Tubuai), primarily due to false positives from proteins, cellulose, and stearates. In Tubuai, polyamide (28.9%), PVC (12.6%), and polystyrene (10.7%) were the dominant polymers, suggesting contributions from fishing gear, agriculture, and household waste. While NR-based quantification overestimates absolute MP counts, the automated pipeline demonstrates potential for high-throughput image processing, reproducible sample analysis, and methodological standardization. This workflow represents a first step toward scalable, low-cost approaches for MPs monitoring in insular systems, highlighting areas for further calibration and optimization. Future work should refine fluorescence thresholds and expand validation across species and locations.
SN - 1932-6203
ER -