Feature Engineering with Microarray Gene Expression Techniques for Asymptomatic Disease Classification

The microarray gene expression patterns (MGEP) allow gene microarray technology to classify cancers at a very early stage. Thus, cancer categorization gene expression profile studies can improve Computer Aided Diagnosis (CAD) systems. However, designing the CAD using MGEP is a challenging research problem. This paper presents a systematic analysis of the present solutions for processing MGEP for disease identification. The recent studies mainly worked on complex MGEP data, pre-processing, important feature selection, and classification. State-of-the-arts were explored followed by the investigation of commonly used MGEP feature selection methods using publicly available datasets. This work designs and investigates feature selection approaches such as Fisher score (FS), Relief F (RF), Signal Noise Ratio (SNR), and Information gain (IG). We used Artificial Neural Network (ANN), Support Vector Machine (SVM), Naive Bayes (NB), and Adaboost classifiers to analyze the results. The outcome of this paper demonstrates the different challenges of existing methods.