Classification of pathogenic bacteria by Raman spectroscopy combined with variational auto‐encoder and deep learning

Abstract

Rapid and early identification of pathogens is critical to guide antibiotic therapy. Raman spectroscopy as a noninvasive diagnostic technique provides rapid and accurate detection of pathogens. Raman spectrum of single cells serves as the “fingerprint” of the cell, revealing its metabolic characteristics. Rapid identification of pathogens can be achieved by combining Raman spectroscopy and deep learning. Traditional classification techniques frequently require lots of data for training, which is time costing to collect Raman spectra. For trace samples and strains that are difficult to culture, it is difficult to provide an accurate classification model. In order to reduce the number of samples collected and improve the accuracy of the classification model, a new pathogen detection method integrating Raman spectroscopy, variational auto‐encoder (VAE), and long short‐term memory network (LSTM) is proposed in this paper. We collect the Raman signals of pathogens and input them to VAE for training. VAE will generate a large number of Raman spectral data that cannot be distinguished from the real spectrum, and the signal‐to‐noise ratio is higher than that of the real spectrum. These spectra are input into the LSTM together with the real spectrum for training, and a good classification model is obtained. The results of the experiments reveal that this method not only improves the average accuracy of pathogen classification to 96.9% but also reduces the number of Raman spectra collected from 1000 to 200. With this technology, the number of Raman spectra collected can be greatly reduced, so that strains that are difficult to culture or trace can be rapidly identified.

A Comparative Analysis of Data Synthesis Techniques to Improve Classification Accuracy of Raman Spectroscopy Data

“…In a similar study, Liu et al 19 combined a VAE and Long Short-Term Memory (LSTM) ( 20 ) network for bacterial pathogen classification. Five bacterial pathogens are used to train the VAE for synthetic data generation.…”

Evaluation of Raman spectroscopy combined with the gated recurrent unit serum detection method in early screening of gastrointestinal cancer

“…To overcome this issue, long short-term memory (LSTM) neural networks have been used to selectively maintain or discard certain information from previous stages using three control gates and additive iteration to circumvent the problem of gradient explosion. 23 However, the drawback of LSTM is that its computational complexity is more than three times that of the original RNN. Therefore, we adopted a GRU with the same functionality and basic structure as LSTM, reducing computational complexity.…”

Deep learning‐based ultra‐fast identification of Raman spectra with low signal‐to‐noise ratio

“…The mainstream generative models commonly used are variational self-encoder (VAE), generative adversarial network (GAN), and flow-based generative models. VAE stores potential attributes as probability distributions, but its method of recovering data distribution and calculating loss function can result in a vague generated output [14]. GAN is a deep learning-based generative model that can generate new content, but it only distinguishes between "real" and "fake" images, which can lead to generated images without actual objects but with similar-looking styles [15,16].…”

“…LSTM adds a forgetting mechanism to RNN, making it possible to selectively retain or forget certain data from the previous period. It also avoids the problem of gradient explosion by using an additive method instead of the multiplicative iteration used in RNN [14].…”