Detection of microplastics in fish using computed tomography and deep learning.
| Autor: | Strafella P; CNR IRBIM, Ancona, Italy.; DIISM, Polytechnic University of Marche, Ancona, Italy., Giulietti N; DIII, University of Pavia, Italy., Caputo A; DIISM, Polytechnic University of Marche, Ancona, Italy., Pandarese G; DIISM, Polytechnic University of Marche, Ancona, Italy., Castellini P; DIISM, Polytechnic University of Marche, Ancona, Italy. |
|---|---|
| Jazyk: | angličtina |
| Zdroj: | Heliyon [Heliyon] 2024 Oct 29; Vol. 10 (21), pp. e39875. Date of Electronic Publication: 2024 Oct 29 (Print Publication: 2024). |
| DOI: | 10.1016/j.heliyon.2024.e39875 |
| Abstrakt: | Marine organisms have been observed ingesting microplastic particles, with field analyses indicating fibers and fragments as prevalent forms. Current microplastic detection methods are mainly time-consuming, susceptible to cross-contamination, and expensive. Furthermore, these techniques, being disruptive, do not allow for the exact localization of the microplastic in the sample. This study proposes a new approach using Computed Tomography (CT scan) and Artificial Intelligence for the automatic and non-destructive detection of microplastics in fishes and other species based on the combination of several factors, such as density and shape. The advantages of this methodology include accurate identification of plastic localization, a low risk of cross-contamination, rapid processing, automatic tomographic measurement, efficient data processing, cost-effectiveness, and a high cost-benefit ratio. The herein results highlight how artificial intelligence applied to conventional techniques can significantly improve precision and efficiency in microplastic research. Indeed, the semantic segmentation model clearly recognized the presence of 100 % of the plastic particles, both in their location and in their volume, accelerating the identification process and surpassing the limitations of traditional spectral analysis methodologies. Competing Interests: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. (© 2024 The Authors. Published by Elsevier Ltd.) |
| Databáze: | MEDLINE |
| Externí odkaz: |
|