Learning Diatoms Classification from a Dry Test Slide by Holographic Microscopy

Abstract

Diatoms are among the dominant phytoplankters in marine and freshwater habitats, and important biomarkers of water quality, making their identification and classification one of the current challenges for environmental monitoring. To date, taxonomy of the species populating a water column is still conducted by marine biologists on the basis of their own experience. On the other hand, deep learning is recognized as the elective technique for solving image classification problems. However, a large amount of training data is usually needed, thus requiring the synthetic enlargement of the dataset through data augmentation. In the case of microalgae, the large variety of species that populate the marine environments makes it arduous to perform an exhaustive training that considers all the possible classes. However, commercial test slides containing one diatom element per class fixed in between two glasses are available on the market. These are usually prepared by expert diatomists for taxonomy purposes, thus constituting libraries of the populations that can be found in oceans. Here we show that such test slides are very useful for training accurate deep Convolutional Neural Networks (CNNs). We demonstrate the successful classification of diatoms based on a proper CNNs ensemble and a fully augmented dataset, i.e., creation starting from one single image per class available from a commercial glass slide containing 50 fixed species in a dry setting. This approach avoids the time-consuming steps of water sampling and labeling by skilled marine biologists. To accomplish this goal, we exploit the holographic imaging modality, which permits the accessing of a quantitative phase-contrast maps and a posteriori flexible refocusing due to its intrinsic 3D imaging capability. The network model is then validated by using holographic recordings of live diatoms imaged in water samples i.e., in their natural wet environmental condition.

Keywords: classification; deep learning; diatoms; digital holography; environmental monitoring; marine pollution; microplankton; phase-contrast microscopy; taxonomy; water quality sensors.

Figure 1

Experimental setup. FC: Fiber coupler; OF: Optical Fiber; BC: Beam Combiner; M: Mirror; MO: Microscope Objective; L: lens.

Figure 2

Augmentation of Holographic data provides 174.636 phase-contrast images from one single hologram of the object.

Figure 3

Holographic recording and reconstructions of diatoms within the glass slide. (a) Bright field image of all diatoms on the glass slide (5× commercial microscope). (b,d) are two recorded digital holograms within the red and green Field of View (FoV), respectively, and (c,e) are the corresponding wrapped quantitative phase images (WQPIs) reconstructions.

Figure 4

Initial guess for creating the training dataset. (a) WQPIs of each diatom in the test glass slide, labeled from 1 to 50. (b,f) are two WQPIs selected among the others, on which a cascade of transformations are applied, i.e., resizing (c,g), rotation (d,h) and phase biasing (e,i).

Figure 5

Examples of holographic images of live diatoms. (a) one of the recorded digital holograms of diatoms mixed in a petri dish. (b) class 27 (c) class 41 (d) class 42. Each class correspond to diatoms species. (b–d) Phase-contrast map are shown. Diatoms belonging to these three classes have similar morphological features and are used to carry out the tests.

Figure 6

Confusion matrices related to ensemble predictions. (a) All output predictions. (b) Considering only classes belonging to the test dataset.