Raman spectrum classification based on transfer learning by a convolutional neural network: Application to pesticide detection

MURDA: Multisource Unsupervised Raman Spectroscopy Domain Adaptation Model with Reconstructed Target Domains for Medical Diagnosis Assistance

Artificial intelligence combined with Raman spectroscopy for disease diagnosis is on the rise. However, these methods require a large amount of annotated spectral data for modeling to achieve high diagnostic accuracy. Annotating labels consumes significant medical resources and time. To reduce dependence on labeled medical data resources, we propose a method called Multisource Unsupervised Raman Spectroscopy Domain Adaptation Model with Reconstructed Target Domains (MURDA). It transfers knowledge learned from source domain data sets of different diseases to an unlabeled target domain data set. Compared to knowledge transfer from a single source domain, knowledge from multiple disease source domains provides more generalized knowledge. Considering the diversity of autoimmune diseases and the limited sample size, we apply MURDA to assist in the medical diagnosis of autoimmune diseases. Additionally, we propose a Double-Branch Multiscale Convolutional Self-Attention (DMCS) feature extractor that is more suitable for spectral data feature extraction. On three sets of serum Raman spectroscopy data sets for autoimmune diseases, the multisource domain adaptation diagnostic accuracy of MURDA was superior to traditional single source and multisource models, with accuracy rates of 73.6%, 83.4%, and 82.9%, respectively. Compared with pure source tasks without domain adaptation, it improved by 15.1%, 36%, and 21.6%, respectively. This demonstrates the effectiveness of Raman spectroscopy combined with MURDA in diagnosing autoimmune diseases. We investigated the important decision dependency peaks in migration tasks, providing assistance for future research on artificial intelligence combined with Raman spectroscopy for diagnosing autoimmune diseases. Furthermore, to validate the effectiveness and generalization performance of MURDA, we conducted experiments on the publicly available RRUFF data set, exploring the application of multisource unsupervised domain adaptation in more Raman spectroscopy scenarios.

Determination of Trace Organic Contaminant Concentration via Machine Classification of Surface-Enhanced Raman Spectra

Surface-enhanced Raman spectroscopy (SERS) has been well explored as a highly effective characterization technique that is capable of chemical pollutant detection and identification at very low concentrations. Machine learning has been previously used to identify compounds based on SERS spectral data. However, utilization of SERS to quantify concentrations, with or without machine learning, has been difficult due to the spectral intensity being sensitive to confounding factors such as the substrate parameters, orientation of the analyte, and sample preparation technique. Here, we demonstrate an approach for predicting the concentration of sample pollutants from SERS spectra using machine learning. Frequency domain transform methods, including the Fourier and Walsh-Hadamard transforms, are applied to spectral data sets of three analytes (rhodamine 6G, chlorpyrifos, and triclosan), which are then used to train machine learning algorithms. Using standard machine learning models, the concentration of the sample pollutants is predicted with >80% cross-validation accuracy from raw SERS data. A cross-validation accuracy of 85% was achieved using deep learning for a moderately sized data set (∼100 spectra), and 70-80% was achieved for small data sets (∼50 spectra). Performance can be maintained within this range even when combining various sample preparation techniques and environmental media interference. Additionally, as a spectral pretreatment, the Fourier and Hadamard transforms are shown to consistently improve prediction accuracy across multiple data sets. Finally, standard models were shown to accurately identify characteristic peaks of compounds via analysis of their importance scores, further verifying their predictive value.

Recent advances of biocompatible optical nanobiosensors in liquid biopsy: towards early non-invasive diagnosis

Liquid biopsy is a non-invasive diagnostic method that can reduce the risk of complications and offers exceptional benefits in the dynamic monitoring and acquisition of heterogeneous cell population information. Optical nanomaterials with excellent light absorption, luminescence, and photoelectrochemical properties have accelerated the development of liquid biopsy technologies. Owing to the unique size effect of optical nanomaterials, their improved optical properties enable them to exhibit good sensitivity and specificity for mitigating signal interference from various molecules in body fluids. Nanomaterials with biocompatible and optical sensing properties play a crucial role in advancing the maturity and diversification of liquid biopsy technologies. This article offers a comprehensive review of recent advanced liquid biopsy technologies that utilize novel biocompatible optical nanomaterials, including fluorescence, colorimetric, photoelectrochemical, and Raman broad-spectrum-based biosensors. We focused on liquid biopsy for the most significant early biomarkers in clinical medicine, and specifically reviewed reports on the effectiveness of optical nanosensing technology in the detection of real patient samples, which may provide basic evidence for the transition of optical nanosensing technology from engineering design to clinical practice. Furthermore, we introduced the integration of optical nanosensing-based liquid biopsy with modern devices, such as smartphones, to demonstrate the potential of the technology in portable clinical diagnosis.

Combined Mutual Learning Net for Raman Spectral Microbial Strain Identification

Infectious diseases pose a significant threat to global health, yet traditional microbiological identification methods suffer from drawbacks, such as high costs and long processing times. Raman spectroscopy, a label-free and noninvasive technique, provides rich chemical information and has tremendous potential in fast microbial diagnoses. Here, we propose a novel Combined Mutual Learning Net that precisely identifies microbial subspecies. It demonstrated an average identification accuracy of 87.96% in an open-access data set with thirty microbial strains, representing a 5.76% improvement. 50% of the microbial subspecies accuracies were elevated by 1% to 46%, especially for E. coli 2 improved from 31% to 77%. Furthermore, it achieved a remarkable subspecies accuracy of 92.4% in the custom-built fiber-optical tweezers Raman spectroscopy system, which collects Raman spectra at a single-cell level. This advancement demonstrates the effectiveness of this method in microbial subspecies identification, offering a promising solution for microbiology diagnosis.

The application of laser‑induced fluorescence in oil spill detection

Over the past two decades, oil spills have been one of the most serious ecological disasters, causing massive damage to the aquatic and terrestrial ecosystems as well as the socio-economy. In view of this situation, several methods have been developed and utilized to analyze oil samples. Among these methods, laser-induced fluorescence (LIF) technology has been widely used in oil spill detection due to its classification method, which is based on the fluorescence characteristics of chemical material in oil. This review systematically summarized the LIF technology from the perspective of excitation wavelength selection and the application of traditional and novel machine learning algorithms to fluorescence spectrum processing, both of which are critical for qualitative and quantitative analysis of oil spills. It can be seen that an appropriate excitation wavelength is indispensable for spectral discrimination due to different kinds of polycyclic aromatic hydrocarbons' (PAHs) compounds in petroleum products. By summarizing some articles related to LIF technology, we discuss the influence of the excitation wavelength on the accuracy of the oil spill detection model and proposed several suggestions on the selection of excitation wavelength. In addition, we introduced some traditional and novel machine learning (ML) algorithms and discussed the strengths and weaknesses of these algorithms and their applicable scenarios. With an appropriate excitation wavelength and data processing algorithm, it is believed that laser-induced fluorescence technology will become an efficient technique for real-time detection and analysis of oil spills.

Deep-Learning-Enabled High-Fidelity Absorbance Spectra from Distorted Dual-Comb Absorption Spectroscopy for Gas Quantification Analysis

Dual-comb absorption spectroscopy has been a promising technique in laser spectroscopy due to its intrinsic advantages over broad spectral coverage, high resolution, high acquisition speed, and frequency accuracy. However, two primary challenges, including etalon effects and complex baseline extraction, still severely hinder its implementation in recovering absorbance spectra and subsequent quantification analysis. In this paper, we propose a deep learning enabled processing framework containing etalon removal and baseline extraction modules to obtain absorbance spectra from distorted dual-comb absorption spectroscopy. The etalon removal module utilizes a typical U-net model, and the baseline extraction module consists of a modified U-net model with physical constraint and an adaptive iteratively reweighted penalized least squares method serving as refinement. The training datasets combine experimental baselines and simulated gas absorption with different concentrations, fully exploiting prior information on gas absorption features from the HITRAN database. In the simulated and experimental test, the CO2 absorbance spectrum covering 25 cm-1 shows high consistency with the HITRAN database, of which the mean absolute error is less than 1% of the maximum absorbance value, and the retrieved concentration has a relative error under 2%, outperforming traditional approaches and indicating the potential practicality of our data processing framework. Hopefully, with a larger network volume and proper datasets, this processing framework can be extended to precise quantification analysis in more comprehensive applications such as atmospheric measurement and industrial monitoring.

Recognition of food images based on transfer learning and ensemble learning

The recognition of food images is of great significance for nutrition monitoring, food retrieval and food recommendation. However, the accuracy of recognition had not been high enough due to the complex background of food images and the characteristics of small inter-class differences and large intra-class differences. To solve these problems, this paper proposed a food image recognition method based on transfer learning and ensemble learning. Firstly, generic image features were extracted by using the convolutional neural network models (VGG19, ResNet50, MobileNet V2, AlexNet) pre-trained on the ImageNet dataset. Secondly, the 4 pre-trained models were transferred to the food image dataset for model fine-tuning. Finally, different basic learner combination strategies were adopted to establish the ensemble model and classify feature information. In this paper, several kinds of experiments were performed to compare the results of food image recognition between single models and ensemble models on food-11 dataset. The experimental results demonstrated that the accuracy of the ensemble model was the highest, reaching 96.88%, which was superior to any base learner. Therefore, the convolutional neural network model based on transfer learning and ensemble learning has strong learning ability and generalization ability, and it is feasible and practical to apply the method to food image recognition.

Library-Based Raman Spectral Identification Using Multi-Input Hybrid ResNet

Raman spectroscopy is widely used for its exceptional identification capabilities in various fields. Traditional methods for target identification using Raman spectroscopy rely on signal correlation with moving windows, requiring data preprocessing that can significantly impact identification performance. In recent years, deep-learning approaches have been proposed to leverage data augmentation techniques, such as baseline and additive noise addition, in order to overcome data scarcity. However, these deep-learning methods are limited to the spectra encountered during training and struggle to handle unseen spectra. To address these limitations, we propose a multi-input hybrid deep-learning model trained with simulated spectral data. By employing simulated spectra, our method tackles the challenges of data scarcity and the handling of unseen spectra encountered in traditional and deep-learning methods. Experimental results demonstrate that our proposed method achieves outstanding identification performance and effectively handles spectra obtained from different Raman spectroscopy systems.

Diagnoses in multiple types of cancer based on serum Raman spectroscopy combined with a convolutional neural network: Gastric cancer, colon cancer, rectal cancer, lung cancer

Cancer is one of the major diseases that seriously threaten human health. Timely screening is beneficial to the cure of cancer. There are some shortcomings in current diagnosis methods, so it is very important to find a low-cost, fast, and nondestructive cancer screening technology. In this study, we demonstrated that serum Raman spectroscopy combined with a convolutional neural network model can be used for the diagnosis of four types of cancer including gastric cancer, colon cancer, rectal cancer, and lung cancer. Raman spectra database containing four types of cancer and healthy controls was established and a one-dimensional convolutional neural network (1D-CNN) was constructed. The classification accuracy of the Raman spectra combined with the 1D-CNN model was 94.5%. A convolutional neural network (CNN) is regarded as a black box, and the learning mechanism of the model is not clear. Therefore, we tried to visualize the CNN features of each convolutional layer in the diagnosis of rectal cancer. Overall, Raman spectroscopy combined with the CNN model is an effective tool that can be used to distinguish different cancer from healthy controls.

Detecting different pesticide residues on Hami melon surface using hyperspectral imaging combined with 1D-CNN and information fusion

Efficient, rapid, and non-destructive detection of pesticide residues in fruits and vegetables is essential for food safety. The visible/near infrared (VNIR) and short-wave infrared (SWIR) hyperspectral imaging (HSI) systems were used to detect different types of pesticide residues on the surface of Hami melon. Taking four pesticides commonly used in Hami melon as the object, the effectiveness of single-band spectral range and information fusion in the classification of different pesticides was compared. The results showed that the classification effect of pesticide residues was better by using the spectral range after information fusion. Then, a custom multi-branch one-dimensional convolutional neural network (1D-CNN) model with the attention mechanism was proposed and compared with the traditional machine learning classification model K-nearest neighbor (KNN) algorithm and random forest (RF). The traditional machine learning classification model accuracy of both models was over 80.00%. However, the classification results using the proposed 1D-CNN were more satisfactory. After the full spectrum data was fused, it was input into the 1D-CNN model, and its accuracy, precision, recall, and F1-score value were 94.00%, 94.06%, 94.00%, and 0.9396, respectively. This study showed that both VNIR and SWIR hyperspectral imaging combined with a classification model could non-destructively detect different pesticide residues on the surface of Hami melon. The classification result using the SWIR spectrum was better than that using the VNIR spectrum, and the classification result using the information fusion spectrum was better than that using SWIR. This study can provide a valuable reference for the non-destructive detection of pesticide residues on the surface of other large, thick-skinned fruits.

Non-Destructive Detection of Different Pesticide Residues on the Surface of Hami Melon Classification Based on tHBA-ELM Algorithm and SWIR Hyperspectral Imaging

In the field of safety detection of fruits and vegetables, how to conduct non-destructive detection of pesticide residues is still a pressing problem to be solved. In response to the high cost and destructive nature of existing chemical detection methods, this study explored the potential of identifying different pesticide residues on Hami melon by short-wave infrared (SWIR) (spectral range of 1000-2500 nm) hyperspectral imaging (HSI) technology combined with machine learning. Firstly, the classification effects of classical classification models, namely extreme learning machine (ELM), support vector machine (SVM), and partial least squares discriminant analysis (PLS-DA) on pesticide residues on Hami melon were compared, ELM was selected as the benchmark model for subsequent optimization. Then, the effects of different preprocessing treatments on ELM were compared and analyzed to determine the most suitable spectral preprocessing treatment. The ELM model optimized by Honey Badger Algorithm (HBA) with adaptive t-distribution mutation strategy (tHBA-ELM) was proposed to improve the detection accuracy for the detection of pesticide residues on Hami melon. The primitive HBA algorithm was optimized by using adaptive t-distribution, which improved the structure of the population and increased the convergence speed. Compared the classification results of tHBA-ELM with HBA-ELM and ELM model optimized by genetic algorithm (GA-ELM), the tHBA-ELM model can accurately identify whether there were pesticide residues and different types of pesticides. The accuracy, precision, sensitivity, and F1-score of the test set was 93.50%, 93.73%, 93.50%, and 0.9355, respectively. Metaheuristic optimization algorithms can improve the classification performance of classical machine learning classification models. Among all the models, the performance of tHBA-ELM was satisfactory. The results indicated that SWIR-HSI coupled with tHBA-ELM can be used for the non-destructive detection of pesticide residues on Hami melon, which provided the theoretical basis and technical reference for the detection of pesticide residues in other fruits and vegetables.

Analyzing the serum of hemodialysis patients with end-stage chronic kidney disease by means of the combination of SERS and machine learning

The aim of this paper is a multivariate analysis of SERS characteristics of serum in hemodialysis patients, which includes constructing classification models (PLS-DA, CNN) by the presence/absence of end-stage chronic kidney disease (CKD) with dialysis and determining the most informative spectral bands for identifying dialysis patients by variable importance distribution. We found the spectral bands that are informative for detecting the hemodialysis patients: the 641 cm^-1, 724 cm^-1, 1094 cm^-1 and 1393 cm^-1 bands are associated with the degree of kidney function inhibition; and the 1001 cm^-1 band is able to demonstrate the distinctive features of hemodialysis patients with end-stage CKD.

Spectral Encoder to Extract the Features of Near-Infrared Spectra for Multivariate Calibration

An autoencoder architecture was adopted for near-infrared (NIR) spectral analysis by extracting the common features in the spectra. Three autoencoder-based networks with different purposes were constructed. First, a spectral encoder was established by training the network with a set of spectra as the input. The features of the spectra can be encoded by the nodes in the bottleneck layer, which in turn can be used to build a sparse and robust model. Second, taking the spectra of one instrument as the input and that of another instrument as the reference output, the common features in both spectra can be obtained in the bottleneck layer. Therefore, in the prediction step, the spectral features of the second can be predicted by taking the reverse of the decoder as the encoder. Furthermore, transfer learning was used to build the model for the spectra of more instruments by fine-tuning the trained network. NIR datasets of plant, wheat, and pharmaceutical tablets measured on multiple instruments were used to test the method. The multi-linear regression (MLR) model with the encoded features was found to have a similar or slightly better performance in prediction compared with the partial least-squares (PLS) model.

Classification of skin cancer using convolutional neural networks analysis of Raman spectra

BACKGROUND AND OBJECTIVE

Skin cancer is the most common malignancy in whites accounting for about one third of all cancers diagnosed per year. Portable Raman spectroscopy setups for skin cancer "optical biopsy" are utilized to detect tumors based on their spectral features caused by the comparative presence of different chemical components. However, low signal-to-noise ratio in such systems may prevent accurate tumors classification. Thus, there is a challenge to develop methods for efficient skin tumors classification.

METHODS

We compare the performance of convolutional neural networks and the projection on latent structures with discriminant analysis for discriminating skin cancer using the analysis of Raman spectra with a high autofluorescence background stimulated by a 785 nm laser. We have registered the spectra of 617 cases of skin neoplasms (615 patients, 70 melanomas, 122 basal cell carcinomas, 12 squamous cell carcinomas and 413 benign tumors) in vivo with a portable Raman setup and created classification models both for convolutional neural networks and projection on latent structures approaches. To check the classification models stability, a 10-fold cross-validation was performed for all created models. To avoid models overfitting, the data was divided into a training set (80% of spectral dataset) and a test set (20% of spectral dataset).

RESULTS

The results for different classification tasks demonstrate that the convolutional neural networks significantly (p<0.01) outperforms the projection on latent structures. For the convolutional neural networks implementation we obtained ROC AUCs of 0.96 (0.94 - 0.97; 95% CI), 0.90 (0.85-0.94; 95% CI), and 0.92 (0.87 - 0.97; 95% CI) for classifying a) malignant vs benign tumors, b) melanomas vs pigmented tumors and c) melanomas vs seborrheic keratosis respectively.

CONCLUSIONS

The performance of the convolutional neural networks classification of skin tumors based on Raman spectra analysis is higher or comparable to the accuracy provided by trained dermatologists. The increased accuracy with the convolutional neural networks implementation is due to a more precise accounting of low intensity Raman bands in the intense autofluorescence background. The achieved high performance of skin tumors classifications with convolutional neural networks analysis opens a possibility for wide implementation of Raman setups in clinical setting.