Non-small cell lung cancer prediction based on orthogonal discrimination projection
Over the past few decades, large families of dimensionality reduction methods such as supervised and unsupervised or linear and non-linear have been developed. In this study, a method of dimensionality reduction for cancer data is presented, which is based on their gene expression signature to specific diagnostic categories of Non-small cell lung cancer. The proposed algorithm, named as Orthogonal Discriminant Projection (ODP), is a linear approximation to manifold learning based approach. The ODP method characterizes the local and non-local information of manifold distributed data and explores an optimum subspace which can maximize the weighted difference between non-local scatter and the local scatter. Moreover, the class information is introduced to enhance the recognition ability. Experimental results on Non-small Cell Lung Cancer (NSCLC) data validates its efficiency compared to other widely used dimensionality reduction methods such as Principle Component Analysis (PCA) and Linear Discriminant Analysis (LDA).