An Compressive Sensing Approach for Subtyping of Gliomaby

An article by admin    No Comments

Gliomas are tumors originating from brain or spine and arise from glial cells. Gliomas are the most common type of primary brain tumors in adults. The classification of gliomas can be based on cell type, grade and location. For instance, gliomas can be classified into
low-grade and high-grade determined by pathologic evaluation of the tumor. In this study Tang et al. we define the subtypes based on genetic and molecular signatures according to the reference.

Combined analysis with different data typesof genome-wide measurements is not a new concept, but how to combine them efficiently for biological discovery is always challenging. A web based platform, called Magellan, was developed for the integrated analysis of DNA copy number and expression data in ovarian cancer. The significant correlation between gene expression and patient survival has been found by Magellan. Troyanskaya et al.  developed a Bayesian framework to combine heterogeneous data sources for predicting gene function, and a improved accuracy of the gene groupings has been achieved compared with microarray analysis alone. However according to Tang et al. glioma’s needed more data types for their classification for improving the accuracy in treatment, and life expectancy. Tang et al. have opted for the novel approach of compressed sensing (CS), also called compressive sampling, has been developed recently in statistics and signal processing, and becomes a powerful tool for data analysis.

Overall Tang et al. prove that the combined analysis method proposed provides an improved way of sub-typing gliomas than using an individual data. It has the potential to improve the diagnostic accuracy in the clinical practice.