Pulmonary tuberculosis is one of the top 10 causes of death in the world in which Latin America represents 3% of the worldwide incidence with a mortality rate of 7.3%. According to goals of World Health Organization to end the global tuberculosis epidemy by 2030, the development of novel diagnosis strategies is crucial. Taking this into account, the present work shows results on the use of Convolutional Neural Networks to aid Computer Vision diagnosis of tuberculosis from patients’ chest X-ray. The proposition of an ensemble using three different deep architectures of such networks shows an important potential of the solution proposed on this task, with accuracy higher than 93%. Such results were obtained from real-world patient data from public datasets, favouring reproducibility, surpasses experts performance reported by literature in these very same datasets, uses canonical architectures of convolutional neural networks, with and without Transfer Learning from different domains, and requires minimal effort on data preparation and no previous feature extraction.