Search

This study presents a novel method for the diagnosis of hepatitis B virus infection using human blood serum Raman spectroscopy combined with a deep learning model. The sera of 499 people infected with the hepatitis B virus and 435 healthy controls were measured in this experiment. The data were subjected to a dimensionality reduction by principal component analysis. Then, the features of multiple scales were preserved and fused by a multiscale fusion convolution operation. The gated recurrent unit network was added to extract time series features and finally output the result of the classification through a softmax layer. A diagnostic model based on a gated recurrent unit and multiscale fusion convolutional neural network was constructed and evaluated by a 10-fold cross-validation method. Compared to existing analysis methods for serum Raman spectroscopy, the proposed model achieved the best performance. The combination of Raman spectroscopy and deep learning models is expected to be applied well in the early screening of hepatitis B and is a promising screening method. [ABSTRACT FROM AUTHOR]

Surface-enhanced Raman spectroscopy (SERS), as a rapid, non-invasive and reliable spectroscopic detection technique, has promising applications in disease screening and diagnosis. In this paper, an annealed silver nanoparticles/porous silicon Bragg reflector (AgNPs/PSB) composite SERS substrate with high sensitivity and strong stability was prepared by immersion plating and heat treatment using porous silicon Bragg reflector (PSB) as the substrate. The substrate combines the five deep learning algorithms of the improved AlexNet, ResNet, SqueezeNet, temporal convolutional network (TCN) and multiscale fusion convolutional neural network (MCNN). We constructed rapid screening models for patients with primary Sjögren's syndrome (pSS) and healthy controls (HC), diabetic nephropathy patients (DN) and healthy controls (HC), respectively. The results showed that the annealed AgNPs/PSB composite SERS substrates performed well in diagnosing. Among them, the MCNN model had the best classification effect in the two groups of experiments, with an accuracy rate of 94.7% and 92.0%, respectively. Previous studies have indicated that the AgNPs/PSB composite SERS substrate, combined with machine learning algorithms, has achieved promising classification results in disease diagnosis. This study shows that SERS technology based on annealed AgNPs/PSB composite substrate combined with deep learning algorithm has a greater developmental prospect and research value in the early identification and screening of immune diseases and chronic kidney disease, providing reference ideas for non-invasive and rapid clinical medical diagnosis of patients. [ABSTRACT FROM AUTHOR]

Considering the poor effect of short text classification due to insufficient semantic information mining in the current short text matching methods, a new short text classification method is proposed based on explicit and implicit multiscale weighting semantic information interaction. First, the explicit and implicit representations of short text are obtained by a word vector model (word2vec), convolutional neural networks (CNNs), and long short-term memory (LSTM). Then, a multiscale convolutional neural network obtains the explicit and implicit multiscale weighting semantics information of short text. Finally, the multiscale weighting semantics is fused for more accurate short text classification. The experimental results show that this method is superior to the existing classical short text classification algorithms and two advanced short text classification models on the five short text classification datasets of MR, Subj, TREC, SST1 and SST2 with accuracies of 85.7%, 96.9%, 98.1%, 53.4% and 91.8%, respectively. [ABSTRACT FROM AUTHOR]

Surface-enhanced Raman spectroscopy (SERS), as a rapid, non-invasive and reliable spectroscopic detection technique, has promising applications in disease screening and diagnosis. In this paper, an annealed silver nanoparticles/porous silicon Bragg reflector (AgNPs/PSB) composite SERS substrate with high sensitivity and strong stability was prepared by immersion plating and heat treatment using porous silicon Bragg reflector (PSB) as the substrate. The substrate combines the five deep learning algorithms of the improved AlexNet, ResNet, SqueezeNet, temporal convolutional network (TCN) and multiscale fusion convolutional neural network (MCNN). We constructed rapid screening models for patients with primary Sjögren's syndrome (pSS) and healthy controls (HC), diabetic nephropathy patients (DN) and healthy controls (HC), respectively. The results showed that the annealed AgNPs/PSB composite SERS substrates performed well in diagnosing. Among them, the MCNN model had the best classification effect in the two groups of experiments, with an accuracy rate of 94.7% and 92.0%, respectively. Previous studies have indicated that the AgNPs/PSB composite SERS substrate, combined with machine learning algorithms, has achieved promising classification results in disease diagnosis. This study shows that SERS technology based on annealed AgNPs/PSB composite substrate combined with deep learning algorithm has a greater developmental prospect and research value in the early identification and screening of immune diseases and chronic kidney disease, providing reference ideas for non-invasive and rapid clinical medical diagnosis of patients. [ABSTRACT FROM AUTHOR]

Fennel contains many antioxidant and antibacterial substances, and it has very important applications in food flavoring and other fields. The kinds and contents of chemical substances in fennel vary from region to region, which can affect the taste and efficacy of the fennel and its derivatives. Therefore, it is of great significance to accurately classify the origin of the fennel. Recently, origin detection methods based on deep networks have shown promising results. However, the existing methods spend a relatively large time cost, a drawback that is fatal for large amounts of data in practical application scenarios. To overcome this limitation, we explore an origin detection method that guarantees faster detection with classification accuracy. This research is the first to use the machine learning algorithm combined with the Fourier transform-near infrared (FT-NIR) spectroscopy to realize the classification and identification of the origin of the fennel. In this experiment, we used Rubberband baseline correction on the FT-NIR spectral data of fennel (Yumen, Gansu and Turpan, Xinjiang), using principal component analysis (PCA) for data dimensionality reduction, and selecting extreme learning machine (ELM), Convolutional Neural Network (CNN), recurrent neural network (RNN), Transformer, generative adversarial networks (GAN) and back propagation neural network (BPNN) classification model of the company realizes the classification of the sample origin. The experimental results show that the classification accuracy of ELM, RNN, Transformer, GAN and BPNN models are above 96%, and the ELM model using the hardlim as the activation function has the best classification effect, with an average accuracy of 100% and a fast classification speed. The average time of 30 experiments is 0.05 s. This research shows the potential of the machine learning algorithm combined with the FT-NIR spectra in the field of food production area classification, and provides an effective means for realizing rapid detection of the food production area, so as to merchants from selling shoddy products as good ones and seeking illegal profits. [ABSTRACT FROM AUTHOR]

With the development of commodity economy, the emergence of fake and shoddy raisin has seriously harmed the interests of consumers and enterprises. To deal with this problem, a classification method combining near-infrared spectroscopy and pattern recognition algorithms were proposed for adulterated raisins. In this study, the experiment was performed by three kinds of raisins in Xinjiang (Hongxiangfei, Manaiti, Munage). After collecting and normalizing the spectral data, we compared the spectra of three kinds of raisins. Next the principal component analysis (PCA) was preformed to compress the dimension of the spectral data, and then classification models including support vector machine (SVM), multiscale fusion convolutional neural network (MCNN) and improved AlexNet were established to identify raisins. The accuracy of SVM, MCNN, and improved AlexNet is 100%, 92.83%, and 97.78% respectively. This study proves that near-infrared spectroscopy combined with pattern recognition is feasible for the raisin inspection. [ABSTRACT FROM AUTHOR]

Medical diagnosis technology based on convolutional neural networks (CNNs) has achieved good performance. In this study, we collected serum samples from 38 glioma patients and 45 healthy controls and used partial least squares (PLS) analysis to reduce the dimension of the data. Different levels of noise were added to the reduced data onto data augmentation, and the AlexNet, ResNet, and GoogLeNet fine‐tuning models were applied for classification. To evaluate the performance of the models, we used five‐fold cross‐validation. The accuracy rates of the AlexNet, ResNet, and GoogLeNet fine‐tuning models were 98.50%, 98.24%, and 99.50%, respectively. The model with the best classification effect was GoogLeNet. The specificity and sensitivity of this model were 98.98% and 98.48%, respectively. In addition, the area under the receiver operating characteristic (ROC) curve (AUC) of the established diagnostic model was 0.9949. The results showed that the combination of serum Raman spectroscopy and the PLS‐Gaussian‐GoogLeNet model achieved a good diagnostic effect for glioma. This method has high clinical application value and is worthy of further popularization. [ABSTRACT FROM AUTHOR]

Zhejiang Suichang native honey, which is included in the list of China's National Geographical Indication Agricultural Products Protection Project, is very popular. This study proposes a method of Raman spectroscopy combined with machine learning algorithms to accurately detect low-concentration adulterated Suichang native honey. In this study, the native honey collected by local beekeepers in Suichang was selected for adulteration detection. The spectral data was compressed by Savitzky–Golay smoothing and partial least squares (PLS) in sequence. The PLS features taken for further analysis were selected according to the contribution rate. In this study, three classification modeling methods including support vector machine, probabilistic neural network and convolutional neural network were adopted to correctly classify pure and adulterated honey samples. The total accuracy was 100%, 100% and 99.75% respectively. The research result shows that Raman spectroscopy combined with machine learning algorithms has great potential in accurately detecting adulteration of low-concentration honey. [ABSTRACT FROM AUTHOR]

In recent years, liver cancer has caused great harm to human health. This study aims to use serum Raman spectroscopy combined with deep learning algorithms to classify liver cancer patients and control groups. For improving the robustness of models, we added equal proportions of Gaussian white noise of 5, 10, 15, 20, 25 dBW to enhance the data, and compared the results of convolutional neural networks and long short-term memory networks which show that the convolutional neural network combined with 10-fold data enhancement was better. The accuracy was 96.95%, indicating the great potential of this experimental model in liver cancer detection. [ABSTRACT FROM AUTHOR]

In this experiment, we collected 45 samples of cervicitis, 29 samples of low‐grade squamous intraepithelial lesion (LSIL), 44 samples of high‐grade squamous intraepithelial lesion (HSIL), 39 samples of cervical squamous cell carcinoma, and 38 cases of cervical adenocarcinoma. After preprocessing of the Raman spectral data, partial least squares (PLS) was used to reduce the dimensionality, and then extreme gradient boosting (XGBoost) was used for feature selection to obtain the first 30‐dimensional features. The preprocessed Raman spectral data also used a fast Fourier transform (FFT) to obtain amplitude information, and then PLS and XGBoost were used to obtain the first 30‐dimensional features. Finally, K nearest neighbor (KNN), extreme learning machine (ELM), artificial bee colony support vector machine (ABC‐SVM), support vector machine optimized by the cuckoo search algorithm (CS‐SVM), particle swarm optimization coupled with support vector machine (PSO‐SVM), and the convolutional neural network combined with long‐ and short‐term memory (CNN‐LSTM) classification models were established. In the raw Raman spectral features experiments, the classification accuracies of KNN, ELM, ABC‐SVM, CS‐SVM, PSO‐SVM, and CNN‐LTSM were 60.76%, 65.81%, 76.21%, 77.66%, 73.50%, and 69.19%, respectively. In the feature fusion experiments, the classification accuracies were 60.91%, 67.84%, 77.64%, 78.49%, 75.54%, and 70.72%, respectively. The experimental results show that feature fusion can further improve model performance regardless of whether using linear classification models or nonlinear classification models. Therefore, it provides a new strategy for extracting features and screening multiple cervical pathological tissues in the future. [ABSTRACT FROM AUTHOR]

Lung cancer and glioma are common malignancies worldwide and pose a serious threat to human health. There may be a certain correlation between lung cancer patients and glioma patients in serum composition, but to date, no study on the classification and correlation of lung cancer and glioma is available. In this paper, the differences and relationships between lung cancer and glioma were analyzed from serum Raman spectra. The existing detection methods of lung cancer and glioma are time consuming and expensive, so we propose a method based on patient serum Raman spectra combined with deep learning, which can screen lung cancer and glioma accurately with speed and low cost. In this study, features were extracted from the original spectral data of patients with lung cancer, glioma, and control subjects. By adding different decibels of white Gaussian noise to the training set for data enhancement, the enhanced training set data were imported into a multilayer perceptron (MLP), recursive neural network (RNN), convolutional neural network (CNN), and AlexNet using fivefold cross‐validation to build the diagnostic model. The results show that PLS‐AlexNet is the best model. The accuracy of this model in the binary classification experiment of lung cancer and control subjects, lung cancer and glioma, and glioma and control subjects were 99%, 95.2%, and 100%, respectively, and the experimental accuracy of the AlexNet triclassification algorithm is also above 85%. This method has great potential in clinical diagnosis of diseases. [ABSTRACT FROM AUTHOR]

Diabetic kidney disease (DKD) is one of the most common kidney diseases worldwide. It is estimated that approximately 537 million adults worldwide have diabetes, and up to 30%–40% of diabetic patients are at risk of developing nephropathy. The pathogenesis of DKD is complex, and its onset is insidious. Currently, the clinical diagnosis of DKD primarily relies on the increase of urinary albumin and the decrease in glomerular filtration rate in diabetic patients. However, the excretion of urinary albumin is influenced by various factors, such as physical activity, infections, fever, and high blood glucose, making it challenging to achieve an objective and accurate diagnosis. Therefore, there is an urgent need to develop an efficient, fast, and low-cost auxiliary diagnostic technology for DKD. In this study, an improved Dual Branch Attention Network (DBAN) was developed to quickly identify DKD. Serum Raman spectroscopy samples were collected from 32 DKD patients and 32 healthy volunteers. The collected data were preprocessed using the adaptive iteratively reweighted penalized least squares (airPLS) algorithm, and the DBAN was used to classify the serum Raman spectroscopy data of DKD. The model consists of a dual branch structure that extracts features using Convolutional Neural Network (CNN) and bottleneck layer modules. The attention module allows the model to learn features specifically, and lateral connections are added between the dual branches to achieve multi-level and multi-scale fusion of shallow and deep features, as well as local and global features, improving the classification accuracy of the experiment. The results of the study showed that compared to traditional deep learning algorithms such as Artificial Neural Network (ANN), CNN, GoogleNet, ResNet, and AlexNet, our proposed DBAN classification model achieved 95.4% accuracy, 98.0% precision, 96.5% sensitivity, and 97.2% specificity, demonstrating the best classification performance. This is the best method for identifying DKD, and has important reference value for the diagnosis of DKD patients, as well as improving the accuracy of medical auxiliary diagnosis. [Display omitted] • A new dual branch attention network (DBAN) is proposed. • The dual branch combines different attention mechanisms to achieve multi-level and multi-scale fusion. • The DBAN model achieves excellent classification performance for diabetic kidney disease diagnosis. [ABSTRACT FROM AUTHOR]

In this paper, a total of 20 sites of single nucleotide polymorphisms (SNPs) on the serotonin 3 receptor A gene (HTR3A) and B gene (HTR3B) are used for feature fusion with age, education and marital status information, and the grid search-support vector machine (GS-SVM), the convolutional neural network (CNN) and the convolutional neural network combined with long and short-term memory (CNN-LSTM) are used to classify and discriminate between alcohol-dependent patients (AD) and the non-alcohol-dependent control group. The results show that 19 SNPs combined with academic qualifications have the best discrimination effect. In the GS-SVM, the area under the receiver operating characteristic (ROC) curve (AUC) is 0.87, the AUC of CNN-LSTM is 0.88, and the performance of the CNN model is the best, with an AUC of 0.92. This study shows that the CNN model can more accurately discriminate AD than the SVM to treat patients in time. [ABSTRACT FROM AUTHOR]

The characteristics of the spectral data are essential for the qualitative analysis of substances. Traditional classification models often need to preprocess the data. However, misuse of preprocessing may change the characteristic information carried by the original data which result in poor model performance. This paper proposes an end‐to‐end deep learning method that combines residual modules to learn features from raw data to improve model performance, which called ResidualSpectra. ResidualSpectra model is compared with three convolutional neural network (CNN) models on the original data. The 15 preprocessing approaches are used to evaluate the preprocessing impact by testing five open‐access mid‐infrared, near‐infrared, and Raman spectra datasets (fruits, meats, olive_oils, Tablets_Nir, Tablets_Raman). In most cases, the ResidualSpectra model performs better than the other three CNN models on five datasets and obtains better results in original data than in preprocessed data. The model is compared with linear discriminant analysis (LDA), nonlinear artificial neural network (ANN), support vector machines (SVM) for original and preprocessed data. The results show that the ResidualSpectra method provides improved results over traditional classification methods in most scenarios. Residual Spectra model is developed to learn features from the original spectra without pre‐processing. This study detailed a deep learning method for qualitative analysis of vibrational spectra. The model includes convolutional layers combined with residual structure to learn features from spectra data. [ABSTRACT FROM AUTHOR]

In recent years, Raman spectroscopy combined with deep learning techniques has been widely used in various fields such as medical, chemical, and geological. However, there is still room for optimization of deep learning techniques and model compression algorithms for processing Raman spectral data. To further optimize deep learning models applied to Raman spectroscopy, in this study time, accuracy, sensitivity, specificity and floating point operations numbers(FLOPs) are used as evaluation metrics to optimize the model, which is named RamanCompact(RamanCMP). The experimental data used in this research are selected from the RRUFF public dataset, which consists of 723 Raman spectroscopy data samples from 10 different mineral categories. In this paper, 1D-EfficientNet adapted to the spectral data as well as 1D-DRSN are proposed to improve the model classification accuracy. To achieve better classification accuracy while optimizing the time parameters, three model compression methods are designed: knowledge distillation using 1D-EfficientNet model as a teacher model to train convolutional neural networks(CNN), proposing a channel conversion method to optimize 1D-DRSN model, and using 1D-DRSN model as a feature extractor in combination with linear discriminant analysis(LDA) model for classification. Compared with the traditional LDA and CNN models, the accuracy of 1D-EfficientNet and 1D-DRSN is improved by more than 20%. The time of the distilled model is reduced by 9680.9s compared with the teacher model 1D-EfficientNet under the condition of losing 2.07% accuracy. The accuracy of the distilled model is improved by 20% compared to the CNN student model while keeping inference efficiency constant. The 1D-DRSN optimized with channel conversion method saves 60% inference time of the original 1D-DRSN model. Feature extraction reduces the inference time of 1D-DRSN model by 93% with 94.48% accuracy. This study innovatively combines lightweight models and model compression algorithms to improve the classification speed of deep learning models in the field of Raman spectroscopy, forming a complete set of analysis methods and laying the foundation for future research. [Display omitted] • Proposed RamanCMP, a Raman spectral classification acceleration method. • Proposed 1D-EfficientNet and 1D-DRSN for spectral data. • Proposed channel conversion method and 1D-DRSN feature extractor. • Testing RamanCMP using the RRUFF public dataset and the external validation dataset for Sjögren's syndrome. [ABSTRACT FROM AUTHOR]

[Display omitted] • In this paper, a new spectral analysis method based on the spectral seal Minkowski distance is proposed. • This paper also presents an auxiliary alcohol dependence diagnosis technique based on Raman spectral structure distance feature and a convolutional neural network for the first time. • We evaluated the good performance of the proposed technique of the study. • The results demonstrate that this study provides a new scheme for auxiliary diagnosis and medical Raman spectroscopy analysis of alcohol dependence. Alcohol dependence is an addictive disease, which can lead to cancer, depression, and other problems. The current diagnostic methods of alcohol dependence are easily disturbed by patients' subjective consciousness, which could lead to wrong diagnoses. To solve this problem, this paper applies serum Raman spectra and a deep learning algorithm to the diagnosis of alcohol dependence and constructs an innovative diagnostic framework. We collected 228 serum samples for spectral analysis, including 119 alcohol-dependent patients and 109 controls. In this framework, spectra are firstly converted to two-dimensional spectrograms by clipping and stitching. Then, a calculating box extracts the distance features between spectral nonadjacent peaks on the spectrogram. Finally, the eigenvalues of each peak are used as pixels of artificial images and input into the ConvNeXt-T neural network model for classification. The sensitivity and specificity of the method are 81.82% and 100%. The results show that it is feasible to use Raman spectroscopy combined with a deep learning algorithm to diagnose alcohol dependence. Compared with directly using the ConvNeXt-T model, the distance feature performs better. [ABSTRACT FROM AUTHOR]

[Display omitted] • For the first time, Raman spectroscopy combined with deep learning is used for the classification and identification of red wine origin. • Corresponding biomarkers identified by Raman spectroscopy of red wine. • Classification of red wine origin by PCA analysis. • Comparative experiment of four deep learning models. • The combination of MCNN and Raman spectroscopy achieves the best results for the classification of red wine origin. In this study, we combined Raman spectroscopy with deep learning for the first time to establish an accurate, simple, and fast method to identify the origin of red wines. We collected Raman spectra from 200 red wine samples of the Cabernet Sauvignon variety from four different origins with a portable Raman spectrometer. The red wine samples, made in 2021, were from the same producer in China. Differences were found by analyzing the Raman spectra of red wine samples. These differences are mainly caused by ethanol, carboxylic acids, and polyphenols. After further analysis, for different origins, the different performances of these substances on the Raman spectrum are related to the climate and geographical conditions of the origin. The Raman spectra were analyzed by principal component analysis (PCA). The data with PCA dimensionality reduction were imported into an artificial neural network (ANN), multifeature fusion convolutional neural network (MCNN), GoogLeNet, and residual neural network (ResNet) to establish red wine origin identification models. The classification results of the model prove that climate, geography, and other conditions can provide support for the classification of red wine origin. The experiments showed that all four models performed well, among which MCNN performed the best with 93.2% classification accuracy, and the area under the curve (AUC) was 0.987. This study provides a new means to classify the origin of red wine and opens up new ideas for identifying origins in the food field. [ABSTRACT FROM AUTHOR]

[Display omitted] • Differences in the biochemical composition of seven types of tissues are analyzed. • A new one-dimensional convolutional hierarchical network (H-CNN) is proposed. • H-CNN classifies tissue samples in a hierarchy from coarse to fine. • H-CNN outperforms traditional classification models. Cervical cancer is one of the most common cancers with a long latent period and slow onset process. Early and accurate identification of the stage of cervical cancer can significantly improve the cure rate and patient survival time. In this study, we collected 699 Raman spectral data of tissue sections from 233 different patients. We analyzed and compared the characteristics and differences of the mean Raman spectra of the seven tissues and pointed out the main differences in the biochemical composition of the seven tissues. In this study, 1D hierarchical convolutional neural network (H-CNN) is proposed by integrating the prior knowledge of hierarchical classification relations with the research of deep learning in Raman spectroscopy. H-CNN is based on CNN and is added with three network branches. Hierarchical classification is performed from coarse to fine for tissue samples of cervicitis, Low-grade Squamous Cell Carcinoma, High-grade Squamous Cell Carcinoma, Well Differentiated Squamous Cell Carcinoma, Moderately Differentiated Squamous Cell Carcinoma, Poorly Differentiated Squamous Cell Carcinoma and cervical adenocarcinoma. To evaluate the recognition performance of H-CNN, we compared it with traditional methods such as Bayesian classifier (NB), decision tree classifier (DT), support vector machine classifier (SVM) and CNN. The experimental results show that H-CNN can accurately identify different classes of tissue sections and has apparent advantages in several aspects such as recognition accuracy, stability and sensitivity compared with the other four traditional recognition methods. The classification Macro-Accuracy of H-CNN can reach 94.91%, Macro-Recall can reach 95.31%, Macro-F1 can reach 95.23%, and Macro-AUC can reach 97.35%. The hierarchical classification method proposed in this study can diagnose patients more accurately. This could lay the foundation for further research on Raman spectroscopy as a clinical diagnostic method for cervical cancer. [ABSTRACT FROM AUTHOR]

[Display omitted] • Triple-negative breast cancer and HER2-positive breast cancer are highly malignant, and early diagnosis is helpful to improve the chances of cure. • Serum Raman spectroscopy combined with deep learning algorithms to diagnose triple-negative breast cancer and HER2-positive breast cancer and achieved great diagnostic results. • The proposed method is helpful for early breast cancer diagnosis and can provide a reference for breast cancer patients to make a personalised diagnosis. Breast cancer is common in women, and its number of patients ranks first among female malignant tumors. Breast cancer is highly heterogeneous, and different types of breast cancer have different biological behaviors and prognoses. Therefore, identifying the different types of breast cancer is of great help in formulating individualized treatment plans. Based on serum Raman spectroscopy and deep learning algorithms, we propose a fast and low-cost diagnosis method for screening triple-negative breast cancer, human epidermal growth factor receptor 2 (HER2)-positive breast cancer, and healthy controls. We collected 75 serum samples in this study, including 23 triple-negative breast cancers, 22 HER2-positive breast cancers, and 30 healthy controls. Using the preprocessed Raman spectra as the input of deep learning, three deep learning models, neural network language model (NNLM), bidirectional long-short-term memory network (BiLSTM), and convolutional neural network (CNN), were established, and the accuracy rates of the three models were 87.78%, 90.37%, and 91.11%, respectively. The experimental results demonstrate the feasibility of serum Raman spectroscopy combined with deep learning algorithms to diagnose breast cancer, which can be used as an effective auxiliary diagnosis method for breast cancer. [ABSTRACT FROM AUTHOR]

[Display omitted] • A novel spectral fusion diagnosis technology is proposed. • A new and efficient MFCNN network is designed to process spectral data. • This study explored the potential of spectral fusion technology for the diagnosis of multiple cancers in this complex clinical environment. According to the limited molecular information reflected by single spectroscopy, and the complementarity of FTIR spectroscopy and Raman spectroscopy, we propose a novel diagnostic technology combining multispectral fusion and deep learning. We used serum samples from 45 healthy controls, 44 non-small cell lung cancer (NSCLC), 38 glioma and 37 esophageal cancer patients, and the Raman spectra and FTIR spectra were collected respectively. Then we performed low-level fusion and feature fusion on the spectral, and used SVM, Convolutional Neural Network-Long-Short Term Memory (CNN-LSTM) and the multi-scale convolutional fusion neural network (MFCNN). The accuracy of low-level fusion and feature fusion models are improved by about 10% compared with single spectral models. [ABSTRACT FROM AUTHOR]

In this study, serum samples from 199 patients with thyroid dysfunction and 183 people with normal thyroid function were collected by Raman spectroscopy, and the data were dimensions-reduced by PCA. The reduced data were input into a multi-feature fusion convolutional neural network (MCNN), the improved AlexNet, VGGNet, GoogLeNet and ResNet, Support Vector Machine (SVM) and Decision Tree (DT) for classification, and the results of the seven classification algorithms were compared. Their classification accuracy are 94.01 %, 91.91 %, 90.34 %, 93.46 %, 92.42 %, 82.78 % and 80.89 %, respectively. The results of this study indicate that the combination of serum Raman spectra and MCNN has a good diagnostic effect for identifying thyroid dysfunction, and it is feasible to improve the classic deep learning models for Raman spectrum classification. [ABSTRACT FROM AUTHOR]