The Feasibility of Automated Identification of Six Algae Types Using Feed-Forward Neural Networks and Fluorescence-Based Spectral-Morphological Features

Abstract

Harmful algae blooms (HABs), which produce lethal toxins, are a growing global concern since they negatively affect the quality of drinking water and have major negative impact on wildlife, the fishing industry, as well as tourism and recreational water use. The goldstandard process employed in the field to identify and enumerate algae requires highly trained professionals to manually observe algae under a microscope, which is a very time-consuming and tedious task. Therefore, an automated approach to identify and enumerate these micro-organisms is much needed. In this study, we investigate the feasibility of leveraging machine learning and fluorescence-based spectral-morphological features to enable the identification of six different algae types in an automated fashion. More specifically, a custom multi-band fluorescence imaging microscope is used to capture fluorescence imaging data of a water sample at six different excitation wavelengths ranging from 405 nm -530 nm. Automated data processing and segmentation was performed on the captured fluorescence imaging data to isolate different micro-organisms from the water sample. A number of morphological and spectral fluorescence features are then extracted from the isolated micro-organism imaging data, and used to train neural network classification models designed for the purpose of identification of the six algae types given an isolated micro-organism. Experimental results using three different neural network classification models (one trained on morphological features, one trained on fluorescencebased spectral features, and one trained on fluorescencebased spectral-morphological features) showed that the use of either fluorescence-based spectral features or fluorescence-based spectral-morphological features to train neural network classification models led to statistically significant improvements in identification accuracy when compared to the use of morphological features (with average identification accuracies of 95.7% ± 3.5% and 96.1% ± 1.5%, respectively). These preliminary results are quite promising, given that the identification accuracy of human taxonomists are typically between the range of 67% and 83%, and thus illustrates the feasibility of leveraging machine learning and fluorescence-based spectral-morphological features as a viable method for automated identification of different algae types.

Total Water Column Analysis Shows the Importance of a Single Species in Subsurface Chlorophyll Maximum Thin Layers in Stratified Waters

“…The present study highlights the importance of a relatively small number of key taxa, so a way of simplifying the task of automated particle recognition would be limiting the number of shapes to be targeted for identification (Davies et al, 2015), which could be done following a brief initial pilot study. Thereafter automated identification could also be improved by the use of neural networks/machine learning (Luo et al, 2018;Deglint et al, 2019;Guo et al, 2021).…”

Hyperspectral Three-Dimensional Fluorescence Imaging Using Snapshot Optical Tomography

“…Using our proposed method, the cornification process could be observed with eight fluorophores simultaneously, and the cell could be reconstructed in 3D, allowing for unprecedented observation information of this process. Furthermore, a recent study [ 40 ] which stressed the importance of automated algae classification, used a fully connected neural network to classify six types of algae by measuring the autofluorescent response from six excitation wavelengths. Since the optical system we built captures the 3D spatial information, as well as the spectral information with high resolution, it is likely possible to use our system along with a similar deep learning model to classify several additional types of algae.…”

Machine Learning and Deep Learning Based Computational Approaches in Automatic Microorganisms Image Recognition: Methodologies, Challenges, and Developments

“…The classifier was trained using back propagation algorithm. Deglint et al [ 83 ] implemented three ANN models for the identification of algae. The study was performed using images of six species of algae.…”

“…= 16 Acc. = 80% Imbalanced dataset Deglint et al [ 83 ] Identification of algae species Binary background-foreground Classifier Shape and fluorescence based spectral features ANN C = 6 TI = 2611 Tr. = 1883 Te.…”