Tuberculosis detection architecture with image processing using the SIFT and K-means algorithm

Calderon Vilca H.D.; Ortega Melgarejo L.M.; Larico Uchamaco G.R.; Cárdenas Mariño F.C. · Computacion y Sistemas (2020) · DOI: 10.13053/cys-24-3-3120

Resumen

Tuberculosis is a lethal disease that attacks the lungs in a similar way to COVID 19, according to the who, until 2018 there were more than 10 million people infected with tuberculosis and 1.5 million died with this disease. Artificial Intelligence algorithms allow detecting these diseases quickly and massively. We present an architecture to detect tuberculosis with image processing on lung radiographs, using the SIFT and K-means algorithms. We have tested the architecture with 300 radiographs, achieving 90.3% accuracy in classification.

Instituciones: Universidad Nacional Mayor de San Marcos; Universidad Nacional Amazónica de Madre de Dios