Optical diagnosis of hepatitis B virus infection in blood plasma using Raman spectroscopy and chemometric techniques

The potential of Raman spectroscopy has been utilized for the diagnosis of hepatitis B virus (HBV) infection in blood plasma. Raman spectra of 24 diseased and 10 healthy samples were used to develop distinct types of support vector machine (SVM) models, including linear, quadratic, and radial basis function (RBF) using multivariate method of principal component analysis (PCA) to reduce the dimensions of the obtained datasets. To assess the diagnostic power of these algorithms, developed models were tested on independent dataset. RBF-based PCA-SVM () model achieved the best performance and yielded accuracy of 98.82%, sensitivity of 98.89%, and specificity of 98.80%. The performance of the SVM models was compared with rerated chemometric method of partial least square regression (PLSR), which has been developed by using the same dataset. The PLSR model attained the diagnostic accuracy of 88%, sensitivity of 93%, and specificity of 78% for same dataset. Our developed model has established promising results compared with state-of-the-art approaches. The results reveal the improved performance of the developed chemometric techniques and clinical prediction potential of HBV by PCA-SVMs in conjunction with Raman spectroscopy.