Metasample-Based Sparse Representation for Tumor Classification

Abstract

A reliable and accurate identification of the type of tumors is crucial to the proper treatment of cancers. In recent years, it has been shown that sparse representation (SR) by 1 l -norm minimization is robust to noise, outliers and even incomplete measurements, and SR has been successfully used for classification. This paper presents a new SR based method for tumor classification using gene expression data. A set of metasamples are extracted from the training samples, and then an input testing sample is represented as the linear combination of these metasamples by l 1 -regularized least square method. Classification is achieved by using a discriminating function defined on the representation coefficients. Since l 1 -norm minimization leads to a sparse solution, the proposed method is called metasample based SR classification (MSRC).Extensive experiments on publicly available gene expression datasets show that MSRC is efficient for tumor classification, achieving higher accuracy than many existing representative schemes.

Metagenes-Based Kernel Sparse Representation for Classification of Tumors

“…Note that, the result of SVM, SRC and MSRC in our experiments are slightly different from those reported in Refs. [2,15,21]. This is probably because the distribution file of cross validation in our experiments is different from them.…”

“…The proposed method compare with several state-of-the art methods, such as the widely used SVM, 2 SRC, 15 MRSC. 21 In our proposed MKSRC method, two parameters should to set, namely, positive regularization parameter and number of metagene of each class p i . The value of is given according to a previous study, 14 as following:…”

“…[15,21] SRC has been successfully applied to tumor classification. The experimentally in these two papers has shown that SRC is quite competitive for classification of tumors.…”

“…Essentially, mathematical operation on gene expression data contributes to highlight particular biological functions, to recover meaningful biological information, to capture alternative structures inherent to the data, to provide biological insight, to reduce noise, and to compress the data in a biologically sensible way. [19][20][21][22] In Ref. [21] there is another viewpoint is that the gene expression pattern can be approximate represented as linear combination of these metagenes.…”

“…[19][20][21][22] In Ref. [21] there is another viewpoint is that the gene expression pattern can be approximate represented as linear combination of these metagenes.…”