Preprocessing Strategies for Sparse Infrared Spectroscopy: A Case Study on Cartilage Diagnostics.

Valeria Tafintseva,Polina Fomina,Vesa Virtanen,Tiril Aurora Lintvedt,Isaac Afara,Boris Zimmermann,Rubina Shaikh,Lassi Rieppo,Boris Mizaikoff,Patrick Krebs,Ervin Nippolainen,Achim Kohler,Hafeez Ur Rehman,Johanne Heitmann Solheim,Simo Saarakkala

doi:10.3390/molecules27030873

Abstract

The aim of the study was to optimize preprocessing of sparse infrared spectral data. The sparse data were obtained by reducing broadband Fourier transform infrared attenuated total reflectance spectra of bovine and human cartilage, as well as of simulated spectral data, comprising several thousand spectral variables into datasets comprising only seven spectral variables. Different preprocessing approaches were compared, including simple baseline correction and normalization procedures, and model-based preprocessing, such as multiplicative signal correction (MSC). The optimal preprocessing was selected based on the quality of classification models established by partial least squares discriminant analysis for discriminating healthy and damaged cartilage samples. The best results for the sparse data were obtained by preprocessing using a baseline offset correction at 1800 cm−1, followed by peak normalization at 850 cm−1 and preprocessing by MSC.

Highlights

Infrared spectroscopy is an emerging technique in biomedical applications that has already demonstrated potential for diagnostics of various pathological conditions, such as cancer, osteoarthritis, and infectious diseases [1,2,3]
The study was done in connection with the Horizon 2020 Research and Innovation Programme (H2020-ICT-2016-2017) project MIRACLE, in which the primary goal was to build a system based on seven quantum cascade lasers (QCLs) lasers in mid-IR region and an ATR probe to assess cartilage quality
In this study we evaluated different preprocessing strategies for such sparse data using spectral data of FTIR-ATR

Summary

Introduction

Infrared spectroscopy is an emerging technique in biomedical applications that has already demonstrated potential for diagnostics of various pathological conditions, such as cancer, osteoarthritis, and infectious diseases [1,2,3]. The method is easy to use, nondestructive and cheap which attracts even more attention to various applications in biomedicine [4,5,6,7] This trend is further accelerated by the development of new photonic devices and light sources in the infrared, such as quantum cascade lasers (QCLs), that are either emitting light on a wide spectrum (frequency comb QCLs), tunable over spectral regions (wavelength-tuning QCLs) or have fixed wavelengths, and light-emitting diodes (LEDs) that cover narrow regions in the mid-infrared [8]. QCLs and LEDs with fixed wavelengths allow fast and relatively inexpensive measurements of samples Their use results in a loss of information, as only part of the broad mid-infrared region can be covered. A number of studies have shown that sparse data is often sufficient for creating good discrimination models in IR spectroscopy [9,10,11,12,13]

Objectives

Methods

Results

Conclusion