Machine learning ellipsometry as a sensitive diagnostic tool to study reproductive biology in Zika virus infected murine models

Paulo H.R. Amaral,Natália Teixeira Wnuk,Vidyleison Neves Camargos,Lídia M. Andrade,M.I.N. da Silva,Mauro Martins Teixeira,Danielle da Glória Souza,Vivian Vasconcelos Costa,Samyra Maria dos Santos Nassif Lacerda,Guilherme Mattos Jardim Costa,J.C. González

DOI: https://doi.org/10.1016/j.microc.2024.111973

IF: 5.304

2024-11-04

Microchemical Journal

Abstract:Machine-learning spectroscopy is a recent scientific and fast-growing area where specific machine-learning solutions are used to process spectroscopic data beyond traditional physical modeling. Here, several machine learning models are optimized and trained for processing highly sensitive broadband variable angle spectroscopic ellipsometry measurements as a new tool in reproductive biology. This new method is applied to classify semen samples obtained from male offspring from ZIKV-infected and uninfected murine models. The method is label-free, cost-effective, simple, and reaches excellent performance. The combination of spectroscopic depolarization degree data with a well-optimized and trained SVM model resulted in the excellent performance of AUROC = 0.99, accuracy = 92%, specificity = 87% and sensibility = 97%. Optical modeling of ellipsometry spectra with Kramers–Kronigconsistent B-splines that takes care of incoherent reflections in the samples allowed for the extraction of the average semen sample refraction index for each class, revealing small differences that were attributed to the observed semen head defects, protamination failure, and DNA fragmentation.

chemistry, analytical