Biomarker Selection, Employing an Iterative Peak Selection Method, and Prostate Spectra Characterization for Identifying Biomarkers Related to Prostate Cancer

A proteomic analysis system (PAS) for prostate Mass Spectrometry (MS) spectra is proposed for differentiating normal from abnormal and benign from malignant cases and for identifying biomarkers related to prostate cancer. PAS comprised two stages, 1/a pre-processing stage, consisting of MS-spectrum smoothing, normalization, iterative peak selection, and peak alignment, and 2/a classification stage, comprising a 2-level hierarchical tree structure, employing the PNN and SVM classifiers at the 1 (normalabnormal) and 2 (benign-malignant) classification levels respectively. PAS first applied local thresholding, for determining the MS-spectrum noise level, and second an iterative global threshold estimation algorithm, for selecting peaks at different intensity ranges. Two optimum sub-sets of these peaks, one at each global threshold, were used to optimally design the hierarchical classification scheme and, thus, indicate the best m/z values. The information rich biomarkers 1160.8, 2082.2, 3595.9, 4275.3, 5817.3, 7653.2, that have been associated with the prostate gland, are proposed for further investigation.