Cancer Classification Using Pattern Recognition and Computer Vision Techniques

\nThe rapid advancement of DNA microarray technology has significantly contributed to the classification of various cancers, particularly leukemia. However, the high-dimensional nature of gene expression data presents challenges such as data noise and irrelevant features, leading to reduced prediction accuracy. This study proposes a novel Hybrid Filter-Wrapper Gene Selection (HFWGS) method that integrates filter-based techniques (Signal-to-Noise Ratio, Correlation Coefficient, and ReliefF) with wrapper-based approaches to enhance feature selection for leukemia classification. Additionally, a Hybrid Statistical-Gene Voting (HSGV) approach was implemented to further refine classification accuracy. A comparative analysis of classifiers, including K-Nearest Neighbors (KNN), Support Vector Machines (SVM), and Linear Discriminant Analysis (LDA), demonstrated that the HFWGS method consistently improved classification performance, achieving 100% accuracy with a reduced subset of genes. The proposed methods provide an efficient framework for optimizing gene selection and improving diagnostic accuracy in leukemia, paving the way for more targeted therapeutic interventions.\n