Clasificación de patologías presentes en la columna vertebral mediante técnicas de máquinas de aprendizaje

Abstract

Introduction: This paper shows the result of research entitled “Study of pathologies present in vertebral column using artificial Intelligence Techniques as support of diagnostic processes”, developed in University of Valle between the years 2016 and 2017. Problem: Studies and analyzes that are carried out on the health conditions of human beings are often invasive, which leads to greater issues. Objective: To provide a method of study from biomechanical attributes of human beings for the detection of pathologies present in vertebral column. Methodology: The study was based on testing three pattern recognition techniques, Bayes as a classic recognition technique, and intelligent techniques such as Radial Basis Functions Neural Networks (RBF), Support Vector Machines (SVM) and Probabilistic Neural Networks (PNN). Results: During the classification process of the pathologies to study, the best results were obtained using PNN, while the other ones presented good classification results for a particular pathology. Conclusion: It was proven that study techniques contributes important characteristics to diagnosis processes of pathologies present in the vertebral column, such as disk hernia and spondylolisthesis. Originality: This study was carried out with information from real patients, providing study techniques and important results on the diagnosis of vertebral column pathologies. Limitations: The study of vertebral column pathologies requires more information about the biomechanical attributes of human beings.