Gene selection in microarray data analysis for brain cancer classification

Abstract

Cancer classification has been one of the most challenging tasks in clinical diagnosis. At present cancer classification is done mainly by looking through the cells' morphological differences, which do not always give a clear distinction of cancer subtypes. Unfortunately, this may have a significant impact on the final outcome of whether a patient could be cured effectively. Microarray technology can play an important role on diagnosing which type of disease one is carrying. The gene selection process is critical for developing gene markers for faster and more accurate diagnosis. In this paper, we develop a method using pairwise data comparisons instead of the one-over-the-rest approach used nowadays. Results are evaluated using available clustering techniques including hierarchical clustering and k-means clustering. Using pairwise comparison, the best accuracy achieved is 95% while it is only 83% when using one-over-the-rest approach. ©2006 IEEE.