Application of a Computational Method for the Early Diagnosis of Prostate Cancer Using Proteomic Pattern Recognition

Abstract

This paper presents a method based on the recognition of proteomic patterns for the early diagnosis of prostate cancer, using computational techniques, applied in the database of SELDI-TOF proteomic patterns. The method is based on classifying the individual as to the portability stage of prostate cancer. To do so, the Independent Component Analysis (ICA) technique is used to extract the characteristics, after which are utilized the algorithm of Maximum Relevance and Minimum Redundancy to reduce the computational cost, and finally the Support Vector Machine to obtain the classification. The best result of the method was obtained with a vector of 27 characteristics, achieving accuracy, specificity and sensitivity, respectively of 89.21%, 83.68% and 95.08%.

Keywords

• Independent component analysis
• Prostate cancer
• Support vector machine
• Maximum relevance and minimum redundancy
• Pattern Recognition