Analysis of Raman Spectra by Using Deep Learning Methods in the Identification of Marine Pathogens

Abstract

The need for efficient and accurate identification of pathogens in seafood and the environment has become increasingly urgent, given the current global pandemic. Traditional methods are not only time consuming but also lead to sample wastage. Here, we have proposed two new methods that involve Raman spectroscopy combined with a long short-term memory (LSTM) neural network and compared them with a method using a normal convolutional neural network (CNN). We used eight strains isolated from the marine organism Urechis unicinctus, including four kinds of pathogens. After the models were configured and trained, the LSTM methods that we proposed achieved average isolation-level accuracies exceeding 94%, not only meeting the requirement for identification but also indicating that the proposed methods were faster and more accurate than the normal CNN models. Finally, through a computational approach, we designed a loss function to explore the mechanism reflected by the Raman data, finding the Raman segments that most likely exhibited the characteristics of nucleic acids. These novel experimental results provide insights for developing additional deep learning methods to accurately analyze complex Raman data.

Raman Microscopic Identification of Microorganisms on Metal Surfaces via Support Vector Machines

“…In their 2012 publication, Stöckel et al showed an apparent correspondence between relationships established by Raman spectroscopy and phylogenetic or taxonomic relationships of Bacillus species [ 18 ]. Predictions of genetic relatedness are also possible via Raman spectroscopy and machine learning [ 19 ]. However, the phylogenetic relationship of some species may not always match Raman spectroscopic similarity.…”

Development of deep learning algorithms to discriminate giant cell tumors of bone from adjacent normal tissues by confocal Raman spectroscopy

“…54,55 Particularly, LSTM networks are the most effective solution to sequence learning 56 and have been recently applied to analyzing Raman spectral data. 25,57,58 LSTM networks are an improved version of RNNs, which are explicitly designed to avoid the long-term dependency problem. LSTM generally comprises four main gates, i.e., the input gate, the forget gate, the output gate and the cell state.…”

Extreme Point Sort Transformation Combined With a Long Short-Term Memory Network Algorithm for the Raman-Based Identification of Therapeutic Monoclonal Antibodies

“…Bidirectional input flows in two directions, making a Bi-LSTM network preserve the future and past information. In a previous study, 1,200 data points of the Raman spectrum were directly input into the LSTM network, which caused 800 epochs of training ( Yu et al, 2021 ). Excessive training epochs will lead to overfitting of the model.…”

“…A long short-term memory (LSTM) network is a novel recurrent neural network (RNN) architecture in conjunction with an appropriate gradient-based learning algorithm ( Yu et al, 2021 ). Unlike RNN networks, the LSTM network has a hidden inner state that enables applying the previous sequence input to a later calculation and avoids gradient disappearance through a gating unit system.…”

“…Some studies are currently attempting to use LSTM networks to analyse Raman spectra. Yu et al presented Raman spectroscopy combined with an LSTM network to predict eight strains isolated from the marine organism Urechis unicinctus and compared them with a method using a normal convolutional neural network (CNN) ( Yu et al, 2021 ). The proposed LSTM methods achieved average isolation-level accuracies greater than 94%, which is higher than that of the normal CNN network.…”