Periodontitis detection using Raman spectroscopy, support vector machine, and salivary biomarkers

Abstract

Many research areas have developed techniques to diagnose lung cancer, cardiovascular diseases, stress, caries, and periodontitis by analyzing saliva. This paper describes a study that predicts periodontitis based on Raman spectra of saliva and biomarkers, such as albumin and alanine aminotransferase (ALT). The spectra were smoothed using a Whittaker filter and baseline correction in MATLAB. In addition, a residual analysis of intensities was performed, and the root mean square deviation was calculated and used as a threshold to establish the active bands of interest, based on the Raman bands associated with albumin and ALT. ORCA quantum chemistry software was used to predict the fundamental frequencies and intensities of some saliva constituents. Support vector machines were used to perform spectral distinction and discriminate between healthy and periodontitis patients.

Matching Pursuit for Denoising Raman Spectra, Based on Genetic Algorithm and Hermite Atoms

“…7 The Raman technique uses a single frequency of radiation to focus on the sample, and the detected scattered intensity is proportional to a frequency shift, generating three types of bands, the first with a frequency of light scattering equal to the incident (Rayleigh) and the others as subtraction (Stokes Raman) and addition (anti-Stokes Raman) of these frequencies. 8 This technique has been used in studies of body fluids from blood glucose measurement, 9,10 detection of metabolites in urine [11][12][13] and saliva, 14,15 to the less conventional such as the characterization of human tears. 16,17 Another characteristic of Raman spectroscopy is the large amount of information that is extracted in a single measurement; however, a crucial problem for spectroscopists is the appearance of noise in the acquired Raman spectra, which is an inherent barrier that limits subsequent tasks such as quantification or classification.…”

“…7 The Raman technique uses a single frequency of radiation to focus on the sample, and the detected scattered intensity is proportional to a frequency shift, generating three types of bands, the first with a frequency of light scattering equal to the incident (Rayleigh) and the others as subtraction (Stokes Raman) and addition (anti-Stokes Raman) of these frequencies. 8 This technique has been used in studies of body fluids from blood glucose measurement, 9,10 detection of metabolites in urine 11–13 and saliva, 14,15 to the less conventional such as the characterization of human tears. 16,17…”