A Projection Algorithm to Detect Cancer Using Microarray

The projection algorithm to classify tissues with a large number of genes and a small number of microarrays is proposed. The algorithm is based on the angle formed by two vectors in the n-dimensional space, and takes advantages of the geometrical projection principle. The properties of known tissues can be used to train the algorithm and distinguish between the cancer and normal gene expressions. The gene’s percentiles from an independent data set can be used to create a third vector, which is projected into the previously trained vectors to classify the third vector in one of the two populations, cancer or normal population. The proposed algorithm was implemented to detect cervical cancer in a microarray data set, which contains 8 normal and 25 cancerous tissues, which were randomly selected one thousand of times using a combinatory strategy. The algorithm was compared with three existing algorithms that have been used to solve the microarray classification problem: Fisher discriminate function, logistic regression, and artificial neural networks. Results show that the proposed algorithm outperformed the selected algorithms