Discrimination of elemental responsiveness to tumor chemotherapy by laser-induced breakdown spectroscopy coupled with chemometric methods

Abstract

It is significant to develop novel methods to diagnose the tumor status throughout chemotherapy. In the present work, we focused on identifying the elemental biomarkers of chemotherapy-treated and untreated tumor tissues by laser-induced breakdown spectroscopy (LIBS). The unsupervised algorithm as principal component analysis and three supervised algorithms including partial least squares discrimination analysis (PLS-DA), random forest (RF) as well as support vector machine (SVM) were used to develop efficient classification models. The average predictive accuracy was 90.74% via the PLS-DA, 88.89% via RF, and 83.33% via SVM, respectively. The results highlighted the spectral difference between chemotherapy-treated and untreated samples within the range of visible spectra between 300–700 nm. In the meantime, four major elements were found to contribute the classification over the following order: calcium > magnesium = copper > sodium. The results featured the importance of calcium on element-based therapeutic responsiveness biomarker monitoring via a new LIBS-based vision.