Zhuoyong Zhang

An emphatic orthogonal signal correction-support vector machine method for the classification of tissue sections of endometrial carcinoma by near infrared spectroscopy

A new application of emphatic orthogonal signal correction (EOSC) for baseline correction of near infrared spectra from reflectance measurements of tissue sections is introduced. EOSC was evaluated and compared with principal component orthogonal signal correction (PC-OSC) by using support vector machine (SVM) classifiers. In addition, some exemplary synthetic data sets were created to characterize EOSC coupled to SVM for classification. Orthogonal experimental design coupled with analysis of variance (ANOVA) was used to determine the significant parameters for optimization, which were the OSC method and number of components for the model. EOSC combined with the SVM gave better predictions with respect to a larger number of components and was not as susceptible to overfitting the data as the classifier built with PC-OSC data. These results were supported by simulations using synthetic data sets. EOSC is a softer signal correction approach that retains more signal variance which was exploited by the SVM. Classification rates of 93±1% were obtained without orthogonal signal correction with the SVM. PC-OSC and EOSC data gave similar peak prediction accuracies of 94±1%. The key advantages demonstrated by EOSC were its resistance to overfitting, fine-tuning capability or softness, and the retention of spectral features after signal correction.

Data fusion of near-infrared and mid-infrared spectra for identification of rhubarb

Rhubarb has different medicinal efficacy to official rhubarb and may affect the clinical medication safety. In order to guarantee the quality of rhubarb, we established a method to distinguish unofficial rhubarbs. 52 official and unofficial rhubarb samples were analyzed using near-infrared (NIR) spectroscopy and mid-infrared (MIR) spectroscopy for classification. The feature vectors, which were selected by wavelet compression (WC) and interval partial least squares (iPLS) from NIR, MIR spectra, were fused together for identifying rhubarb samples. Partial least squares-discriminant analysis (PLS-DA), soft independent modeling of class analogies (SIMCA), support vector machine (SVM) and artificial neural network (ANN) were compared for classifying rhubarb. The use of data fusion strategies improved the classification model and allowed correct classification of all the samples.

Identification of rhubarbs by using NIR spectrometry and temperature-constrained cascade correlation networks

Temperature-constrained cascade correlation networks (TCCCNs) were used to identify powdered rhubarbs based on their near-infrared spectra. Different network configurations that used multiple network models with single output (Uni-TCCCN) and single networks with multiple outputs (Multi-TCCCN) were compared. Comparative studies were made by using Latin-partitions and leave-one-out cross-validation methods. Results showed that multiple networks with single output predicted generally better than single network with multiple outputs. Better results with TCCCN models were obtained compared with conventional back propagation neural networks (BPNNs). The effects of parameters on correct identification and parameter optimizations were discussed in detail. With optimized neural network training parameters, NIR spectra from powdered rhubarb samples were classified by a TCCCN model with 100% accuracy.

Terahertz time-domain spectroscopy combined with support vector machines and partial least squares-discriminant analysis applied for the diagnosis of cervical carcinoma

Coupled with terahertz time-domain spectroscopy (THz-TDS) technology, the feasibility for the diagnosis of cervical carcinoma using support vector machines (SVM) and partial least squares-discriminant analysis (PLS-DA) had been studied. The terahertz spectra of 52 specimens of cervix were collected. The performance of the preprocessing methods of multiplicative scatter correction (MSC), Savitzky–Golay (SG) smoothing and first derivative, principal component orthogonal signal correction (PC-OSC) and emphatic orthogonal signal correction (EOSC) were investigated for PLS-DA and SVM models. The effects of the different pretreatment methods with respect to classification accuracy were compared. The PLS-DA and SVM models were validated using the bootstrapped Latin-partition method. The SVM and PLS-DA models optimized with the combination of SG first derivative and PC-OSC preprocessing had the best predictive results with classification rates of 94.0% ± 0.4% and 94.0% ± 0.5%, respectively. The proposed procedure proved that terahertz spectroscopy combined with classifiers provides a technology that has potential as a new diagnosis method for cancer tissue.

Locally linear embedding method for dimensionality reduction of tissue sections of endometrial carcinoma by near infrared spectroscopy

Locally linear embedding (LLE) is introduced here as a nonlinear compression method for near infrared reflectance spectra of endometrial tissue sections. The LLE has been evaluated by using support vector machine (SVM) classifiers and the projected difference resolution (PDR) method. Synthetic data sets devised to resemble near-infrared spectra of tissue samples were used to characterize the performance of the LLE. The LLE was compared using principal component compression (PCC) method to evaluate nonlinear and linear compression. For a set of real tissue samples, if the compressed data were not range-scaled prior to SVM classification, the principal component compressed data gave an average prediction rate of 39±2% while the LLE 94±2%; if range-scaled after compression, the LLE and PCC performed evenly, with maximum average prediction values of 94±2% and 93±2%, respectively. The SVM without compression yielded a classification rate of 92±2%. The prediction accuracy was consistent with PDR results. Without the second derivative preprocessing, the classification rates were 90±3%, 89±2%, and 78±2% for the LLE compressed, the PCC, and no compression classifications by the SVM, respectively.

Forward selection radial basis function networks applied to bacterial classification based on MALDI-TOF-MS.

Forward selection improved radial basis function (RBF) network was applied to bacterial classification based on the data obtained by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS). The classification of each bacterium cultured at different time was discussed and the effect of parameters of the RBF network was investigated. The new method involves forward selection to prevent overfitting and generalized cross-validation (GCV) was used as model selection criterion (MSC). The original data was compressed by using wavelet transformation to speed up the network training and reduce the number of variables of the original MS data. The data was normalized prior training and testing a network to define the area the neural network to be trained in, accelerate the training rate, and reduce the range the parameters to be selected in. The one-out-of-n method was used to split the data set of p samples into a training set of size p-1 and a test set of size 1. With the improved method, the classification correctness for the five bacteria discussed in the present paper are 87.5, 69.2, 80, 92.3, and 92.8%, respectively.

A novel DPSO-SVM system for variable interval selection of endometrial tissue sections by near infrared spectroscopy.

A novel method combining a discrete particle swarm optimization (DPSO) with a support vector machine (SVM) was proposed for the variable interval selection of tissue sections of endometrial carcinoma by near infrared spectroscopy. The DPSO-SVM algorithm includes a multi-stage screening. In each screening step, the DPSO was repeated 50 times using random sampling, and the frequencies that the variable intervals were selected among the 50 repeats were used to select the most probable intervals. The variable intervals with high probabilities were selected and further used in the next screening. Finally, the subset of variable intervals with the highest classification rate was considered as the optimal variable intervals. A synthetic data set mimicking the near infrared (NIR) spectra of tissue samples was applied to evaluate the performance of the DPSO-SVM. For the synthetic data, the classification rates were 74.9 ± 0.9% and 100% for the full spectral range and the six variable intervals selected by the DPSO-SVM. For the real endometrial tissue data, the entire spectral data gave an average accuracy of 69.5 ± 0.5%, while the 20 variable intervals gave 98.5 ± 0.3%. The results showed that the informative variables from the NIR spectra could be selected and high classification accuracy was achieved by the proposed approach.

Near infrared spectroscopy combined with least squares support vector machines and fuzzy rule-building expert system applied to diagnosis of endometrial carcinoma

OBJECTIVE The feasibility of early diagnosis of endometrial carcinoma was studied by least squares support vector machines (LS-SVM) and fuzzy rule-building expert system (FuRES) that classified near infrared (NIR) spectra of tissues. METHODS NIR spectra of 77 specimens of endometrium were collected. The spectra were pretreated by principal component orthogonal signal correction (PC-OSC) and direct orthogonal signal correction (DOSC) methods to improve the signal-to-noise ratio (SNR) and remove the influences of background and baseline. The effects of modeling parameters were investigated using bootstrapped Latin-partition methods. RESULTS The optimal LS-SVM model of the PC-OSC pretreatment method successfully classified the samples with prediction accuracies of 96.8±1.4%. CONCLUSIONS The proposed procedure proved to be rapid and convenient, which is suitable to be developed as a non-invasive diagnosis method for cancer tissue.

Quantitative measurements of binary amino acids mixtures in yellow foxtail millet by terahertz time domain spectroscopy.

Terahertz time domain spectroscopy (THz-TDS) combined with chemometrics has been utilized for the qualitative and quantitative analysis of binary mixtures of l-glutamic acid and l-glutamine which have similar chemical structures and properties. The binary mixtures of amino acids were prepared with yellow foxtail millet matrix, substituted for polyethylene (PE) as previously reported. After proper pretreatment of absorption spectra, quantitative analysis was achieved by partial least squares (PLS) and interval partial least squares (iPLS) regressions. The performance of models was evaluated based on the root mean square error of prediction (RMSEP) and correlation coefficient (R(2)) of cross-validations with bootstrapped Latin partitions as criterion. The iPLS yielded better results with low RMSEP (0.39±0.02%, 0.39±0.02%), and higher R(2) values (0.9904, 0.9906) for glutamine and glutamic acid comparing to the conventional PLS models. Multivariate curve resolution alternating least squares (MCR-ALS) was successfully applied for resolution of pure THz spectra and concentration profiles of two amino acids components from mixtures.

THz-TDS combined with a fuzzy rule-building expert system applied to the identification of official rhubarb samples

Terahertz time-domain spectroscopy (THz-TDS) as a new non-destructive testing method has been applied to the identification of 41 official and unofficial rhubarb samples in the present work. The THz time domain spectra of rhubarb samples were preprocessed and then used to establish an identification model by using fuzzy rule-building expert systems (FuRES). The model was validated using a bootstrapped Latin-partitions (BLPs) method with 10 bootstraps and 4 Latin-partitions. The obtained results showed that the model has good predictive ability with respect to the classification accuracy of 94.8 ± 0.5% and 95.2 ± 0.1% by using the preprocessing methods of Savitzky–Golay (S–G) first derivative combined with either one of two orthogonal signal correction (OSC) methods, respectively. The proposed method showed that THz-TDS combined with chemometrics can be used to identify genuine and counterfeit Chinese herbal medicines, as well as official and unofficial rhubarb.