Detection of prostate cancer combining Raman spectroscopy with multilayer perceptron

In the previous research on medical diagnosis of SERS, the most common machine learning method was Principal Component Analysis-Linear Discriminant Analysis (PCA-LDA), but this method had certain limitations and low classification accuracy. Therefore, this study compared surface-enhanced Raman spectroscopy (SERS) of plasma from Prostate Cancer (PC) patients and healthy controls using a deep learning method known as multilayer perceptron (MLP). First of all, the average spectrum and difference spectrum of the two are made to roughly compare the difference between them. Secondly, several Raman positions with obvious characteristics are selected to conduct histogram comparative analysis. Finally, PCA-LDA algorithm and MLP algorithm are respectively used to diagnose the obtained experimental results, and visualization analysis is carried out in combination with figure. The final results showed that although the classification accuracy of PCA-LDA was also good for PC, the diagnostic results of MLP were better than PCA-LDA, and the sensitivity and specificity were also higher than PCA-LDA. This indicated that MLP algorithm combined with SERS improved the accuracy of early detection of prostate cancer and was expected to be expanded in the early screening of other cancers, which had great potential in clinical application.