Measuring similarity and improving stability in biomarker identification methods applied to Fourier-transform infrared (FTIR) spectroscopy

• Published in: Journal of biophotonics Vol. 7; no. 3-4; pp. 254 - 265
• Main Authors: Trevisan, Júlio, Park, Juhyun, Angelov, Plamen P., Ahmadzai, Abdullah A., Gajjar, Ketan, Scott, Andrew D., Carmichael, Paul L., Martin, Francis L.
• Format: Journal Article
• Language: English
• Published: Berlin WILEY-VCH Verlag 01.04.2014; WILEY‐VCH Verlag; Wiley Subscription Services, Inc
• Subjects: Algorithms; Animals; Biomarkers; Biomarkers - chemistry; biospectroscopy; Brain - metabolism; Brain Neoplasms - diagnosis; Brain Neoplasms - metabolism; Classification; computational framework; data mining; Endometriosis - diagnosis; Endometriosis - metabolism; Female; Fourier transforms; Fourier-transform infrared spectroscopy; Humans; Infrared spectroscopy; Mesocricetus; Models, Statistical; Multivariate Analysis; Reproducibility of Results; Similarity; Spectroscopy; Spectroscopy, Fourier Transform Infrared - methods; Stability; statistical data analysis; validation studies; biospectroscopy; computational framework; biomarkers; data mining; Fourier-transform infrared spectroscopy; statistical data analysis; validation studies; classification
• ISSN: 1864-063X; 1864-0648
• DOI: 10.1002/jbio.201300190

FTIR spectroscopy is a powerful diagnostic tool that can also derive biochemical signatures of a wide range of cellular materials, such as cytology, histology, live cells, and biofluids. However, while classification is a well‐established subject, biomarker identification lacks standards and validation of its methods. Validation of biomarker identification methods is difficult because, unlike classification, there is usually no reference biomarker against which to test the biomarkers extracted by a method. In this paper, we propose a framework to assess and improve the stability of biomarkers derived by a method, and to compare biomarkers derived by different method set‐ups and between different methods by means of a proposed “biomarkers similarity index”. (© 2014 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)