Minimally invasive medical diagnosis through Laser-Induced Breakdown Spectroscopy (LIBS) coupled with machine learning

Untargeted spectroscopic methods allow interrogating biomedical samples without prior hypotheses about specific pathways, and therefore they are a powerful diagnostic tool for poorly understood or initially asymptomatic diseases. In this work, we analyzed micro- drops of biological fluids from patients and animal models with Laser-Induced Breakdown Spectroscopy (LIBS) and coupled the spectra with supervised classification algorithms to distinguish healthy and diseased individuals. LIBS is the optical emission spectroscopy of laser-induced plasmas, and it can provide the elemental fingerprinting of a wide variety of samples [1]. In [2], we designed a voting algorithm based on the use of LIBS difference spectra, that we employed as a data-pretreatment and feature-selection method to couple with Quadratic Discriminant Analysis for Alzheimer's Disease diagnosis in living patients.