Hybrid Feature Selection Algorithm mRMR-ICA for Cancer Classification from Microarray Gene Expression Data.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Microarray gene expression data contains redundant information, complicating accurate cancer classification.
• Effective gene selection is crucial for identifying informative markers for cancer diagnosis.

Purpose of the Study:

• To present a novel hybrid feature selection method, mRMR-ICA, for enhanced cancer classification.
• To combine Minimum Redundancy Maximum Relevance (mRMR) with Imperialist Competition Algorithm (ICA) for efficient gene selection.

Main Methods:

• The mRMR-ICA algorithm preprocesses data using mRMR to reduce redundancy, followed by ICA for feature selection.
• Support Vector Machine (SVM) is employed to evaluate classification accuracy, with a fitness function balancing accuracy and gene count.

Main Results:

• mRMR-ICA was evaluated on ten benchmark microarray datasets, demonstrating improved cancer classification accuracy.
• The method effectively identified fewer informative genes compared to original ICA and other evolutionary algorithms.

Conclusions:

• mRMR-ICA efficiently removes redundant genes, leading to better classification performance with fewer selected genes.
• The hybrid approach enhances computational efficiency and shortens calculation time for gene selection in cancer classification.