A new automated radiolarian image acquisition, stacking, processing, segmentation, and identification workflow

Martin Tetard (1) , Ros Marchant (1, 2) , Giuseppe Cortese (3) , Yves Gally (1) , Thibault de Garidel-Thoron , Luc L Beaufort (1)

1 CEREGE - Centre européen de recherche et d'enseignement des géosciences de l'environnement
2 School of Engineering and Robotic
3 GNS Science [Lower Hutt]

Abstract. Identification of microfossils is usually done by expert taxonomists and requires time and a significant amount of systematic knowledge developed over many years. These studies require manual identification of numerous specimens in many samples under a microscope, which is very tedious and time consuming. Furthermore, identification may differ between operators, biasing reproducibility. Recent technological advances in image acquisition, processing, and recognition now enable automated procedures for this process, from microscope image acquisition to taxonomic identification. A new workflow was developed for automated radiolarian image acquisition, stacking, processing, segmentation, and identification. The protocol includes a newly proposed methodology for preparing radiolarian microscopic slides. We mount 8 samples per slide, using a recently developed 3D-printed decanter that enable the random and uniform settling of particles, and minimise the loss of material. Once ready, slides are automatically imaged using a transmitted light microscope. About 4000 specimens per slide (500 per sample) are captured in digital images which include stacking techniques to improve their focus and sharpness. Automated image processing and segmentation is then performed using a custom plugin developed for the ImageJ software. Each individual radiolarian image is automatically classified by a convolutional neural network (CNN) trained on a radiolarian database (currently 17,065 images, corresponding to 112 classes) using the software, ParticleTrieur. The trained CNN has an overall accuracy of about 90 %. The whole procedure, including the image acquisition, stacking, processing, segmentation and recognition, is entirely automated via a LabVIEW interface, and takes approximately 1 hour per sample. Census data count and classified radiolarian images are then automatically exported and saved. This new workflow paves the way for the analysis of long-term, radiolarian-based palaeoclimatic records from siliceous remains-bearing samples.

Domaines: Sciences de l'environnement

Martin Tetard, Ros Marchant, Giuseppe Cortese, Yves Gally, Thibault de Garidel-Thoron, et al.. A new automated radiolarian image acquisition, stacking, processing, segmentation, and identification workflow. 2020. ⟨hal-03016385⟩