Publication detail

Multivariate models for data library transfer in laser spectroscopy

POŘÍZKA, P. KÉPEŠ, E. KAISER, J.

Original Title

Multivariate models for data library transfer in laser spectroscopy

Type

conference proceedings

Language

English

Original Abstract

czLaser-Induced Breakdown Spectroscopy (LIBS) serves as an exceptional platform for a fastsample analysis. Detected laser-induced plasma (LIP) spectrum is composed of uniqueinformation (i.e. chemical fingerprint) characterizing the sample from which it originates. Atypical LIBS analysis, obtained with applying repetition rate, results in a great number ofmeasurements (spectra). The size of the data matrix is intensified by the number of variablescarried by each LIP spectrum. Therefore, the processing of given data matrix calls forsophisticated statistical algorithms. Chemometric (or Multivariate Data Analysis; MVDA)algorithms are becoming an inevitable part of the spectroscopic analysis [1]. A successfulMVDA implementation is based on the well-balanced data pre-processing (outlier filtering,signal standardization, variable down-selection, etc.). First and the most common step is thedimensionality reduction and data visualization. This is consecutively followed by the modelcreation, further classification, and quantitative analysis. Nevertheless, data and relatedstatistical models are not transferable among LIBS instruments. This is the main drawbackof LIBS which limits its spread over applications (regardless of the great success of the Marsrover, Curiosity). We strive to deliver a feasibility study on the possibility of transfer of datalibraries between LIBS systems. For this, we have created a novel multivariate algorithm thatenables to alter the signal response of individual LIBS systems and to unify their respectivedata models, as schematically depicted in Figure 1.

Keywords

Laser-induced breakdown spectroscopy; chemometrics

Authors

POŘÍZKA, P.; KÉPEŠ, E.; KAISER, J.

Released

9. 9. 2019

Publisher

Spektroskopická společnost Jana Marka Marci

Location

Ioannes Marcus Marci Spectroscopic Society Ke Karlovu 2027/3, 120 00 Praha 2 - Nové Město Czech Republic

ISBN

978-80-88195-13-9

Book

EMSLIBS 2019 Book of abstracts

Edition number

1

Pages from

197

Pages to

198

Pages count

2

URL

BibTex

@proceedings{BUT161595,
  editor="Pavel {Pořízka} and Erik {Képeš} and Jozef {Kaiser}",
  title="Multivariate models for data library transfer in laser spectroscopy",
  year="2019",
  number="1",
  pages="197--198",
  publisher="Spektroskopická společnost Jana Marka Marci",
  address="Ioannes Marcus Marci Spectroscopic Society
Ke Karlovu 2027/3, 120 00 Praha 2 - Nové Město
Czech Republic",
  isbn="978-80-88195-13-9",
  url="http://libs.ceitec.cz/files/281/213.pdf"
}