Detecting Toxic Mixed Red Tide Algae in Qinhuangdao Using Three-Dimensional Fluorescence Spectroscopy

Red tide algae can have devastating effects on a marine’s aquaculture and human health. This is why it is important for researchers and scientists to be able to detect the presence of specific types of red tide algae (1). Several coastal regions, such as Qinhuangdao, China, experience recurring red tides comprising various toxic and non-toxic algae species (1).

In a new study, researchers from Yanshan University in Qinhuangdao utilized three-dimensional (3D) fluorescence spectroscopy combined with chemometrics to successfully detect and classify typical toxic mixed red tide algae in the region. The study was published in Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy (1).

The research team employed a fluorescence spectrometer to measure the 3D fluorescence spectra of the mixed red tide algae samples collected from the Qinhuangdao sea area (1). They identified the excitation wavelengths corresponding to the peak positions in the 3D fluorescence spectrum through contour spectrum analysis (1). This information was then used to create new 3D fluorescence spectrum data with selected feature intervals.

Principal component analysis (PCA) was employed to extract the relevant features from the new three-dimensional fluorescence spectrum data. Subsequently, the researchers utilized two classification models: genetic optimization support vector machine (GA-SVM) and particle swarm optimization support vector machine (PSO-SVM), with the extracted feature data and the original data, respectively. By comparing the two feature extraction methods and classification algorithms, the effectiveness of the approach was evaluated.

GA-SVM and PSO-SVM are optimization techniques used in machine learning. In GA-SVM, a genetic algorithm is applied to optimize the parameters of the support vector machine (SVM) model, such as the kernel function and regularization parameters, by mimicking the process of natural evolution. The genetic algorithm utilizes principles of selection, crossover, and mutation to search for the optimal set of parameters that maximizes the classification accuracy. On the other hand, PSO-SVM employs a particle swarm optimization algorithm, inspired by the behavior of a flock of birds or a school of fish, to iteratively update the parameters of the SVM model. It uses a swarm of particles, each representing a potential solution, to explore the parameter space and converge towards the optimal solution based on the best performance achieved by the particles. Both GA-SVM and PSO-SVM are powerful optimization techniques that enhance the performance of SVM models for classification tasks by finding the most suitable parameter configurations.

The results demonstrated that the principal component feature extraction combined with the GA-SVM classification method achieved an impressive classification accuracy of 92.97% for the test set. The optimal excitation wavelengths were determined to be 420 nm, 440 nm, 480 nm, 500 nm, and 580 nm, with emission wavelengths ranging from 650 nm to 750 nm. These findings highlight the feasibility and efficacy of utilizing three-dimensional fluorescence spectrum characteristics and the genetic optimization support vector machine classification method for the accurate identification of toxic mixed red tide algae in the Qinhuangdao sea area.

This breakthrough detection method provides a valuable tool for monitoring and managing red tide phenomena, enabling prompt identification and mitigation of harmful algal blooms. By accurately identifying the specific types of toxic algae, scientists and authorities can implement targeted measures to safeguard marine ecosystems, aquaculture industries, and human health in the Qinhuangdao region.

Reference

(1) Wang, S.-y.; Bi, W.-h.; Li, X.-y.; et al.A detection method of typical toxic mixed red tide algae in Qinhuangdao based on three-dimensional fluorescence spectroscopy. Spectrochimica Acta Part A: Mol. Biomol. Spectrosc. 2023, 298, 122704. DOI: 10.1016/j.saa.2023.122704