Multi-Resolution Analysis Method for IMS Proteomic Data Biomarker Selection and Classification

Abstract

Even though imaging mass spectrometry (IMS) technique is evolving rapidly, its data analysis capability lags behind. Especially with the improving of IMS data resolution, faster and more accurate data analysis algorithms are required. To meet such challenges in IMS data analysis, an effective and efficient algorithm for IMS data biomarker selection and classification using multiresolution (wavelet) analysis method is proposed. We first applied wavelet transform to IMS data de-noising. The idea of wavelet pyramid method for image matching was then applied for biomarker selection, in which Jaccard similarity was used to measure the similarity of wavelet coefficients. Last, the Naive Bayes classifier was used for classification based on feature vectors in terms of wavelet coefficients. Performance of the algorithm was evaluated in real data applications. Experimental results show that this multi-resolution method has advantages of fast computing and accuracy.