Abstract
BackgroundThe World Health Organization (WHO) grade diagnosis of cancer is essential for surgical outcomes and patient treatment. Traditional pathological grading diagnosis depends on dyes or other histological approaches, which are time-consuming (usually 1-2 days), resource-wasting, and labor-intensive. Fourier transform infrared (FTIR) spectroscopy is a rapid and nondestructive technique that has been widely used for detecting the molecular component changes, which relies on the resonant frequencies absorbance of the molecular bonds.MethodsTo overcome the disadvantages of traditional pathological diagnosis, this paper proposed a novel diagnostic method based on FTIR and artificial neural network (ANN). Firstly, the spectra of high- and low-grade human glioma that without dye were collected by FTIR spectrometer, then the raw data preprocessed with baseline correction and amide I (1649 cm-1) normalization before input into the input-layer of the ANN, after the nonlinear conversion of the neurons in the hidden-layers, the categories were presented in the output-layer. Corresponding to the decrease of the loss function, the weights of the net updated continuously, and finally, the optimized model has the power of prediction for new samples. ResultsAfter training on 6225 spectra sourced from 77 glioma patients, the ANN model reached the prediction accuracy, specificity and sensitivity evaluation metrics above 99%, which was much superior to the common classification method of principal component analysis-linear discriminate analysis (PCA-LDA) (the prediction accuracy, specificity and sensitivity are only 87%, 89% and 86%, respectively). Moreover, rather than the lipid range of 2800-3000 cm-1, the ANN learned the fingerprint characteristics of the infrared spectrum to classify the major histopathologic classes of human glioma. Especially, the diagnosis process of the novel method only requires several minutes. Compared to the traditional pathological diagnosis, the efficiency raises almost 500 times.ConclusionsThe infrared range of fingerprint is the major indicator for cancer progression, and the ANN-based diagnosis method can be streamlined, and create a complementary pathway that is independent of the traditional pathology laboratory.
Application of Wavelet Neural Network in Building Settlement Prediction
