Meeting Banner
Abstract #3447

A Statistical Framework for Biomarker Identification using HR-MAS 2D NMR Spectroscopy

Akram Belghith1, Christophe Collet2, Karim Elbayed3, Lucien Rumbach4, Izzie Jacques Namer5, Jean-Paul Armspach6

1University of Strasbourg, LSIIT - CNRS UMR 7005, Strasbourg, Alsace, France; 2University of Strasbourg, LSIIT - CNRS UMR 7005, France; 3University of Strasbourg, Institut de Chimie; 4Neurology Department CHU Minjoz Besancon -France; 5University of Strasbourg, LINC - CNRS FRE 3289 - France; 6University of Strasbourg, LINC - CNRS FRE 3289, France

In this paper, we propose a new scheme to detect and align simultaneously peaks for biomarker identification. The proposed peak detection and alignment approach is based on the use of evidence theory which is well suited to model uncertainty and imprecision that characterize the 2D NMR HRMAS spectra. The peak detection and alignment results will be then used to identify biomarkers present in the biopsy. We particularly show that the use of fuzzy set theory in our biomarker identification scheme achieves consistently high performance compared to the threshold methods.