Visceral Leishmaniasis Detection Using Deep Learning Techniques and Multiple Color Space Bands.

Leishmaniasis is a group of neglected popular parasitic diseases typical of tropical and subtropical countries, and its most serious form is Visceral Leishmaniasis (VL). Every year, it’s 700,000 to 1 million cases are recorded worldwide, leading to the death of 26,000 to 65,000 people. The diagnostic, performed through parasitological examination, is very tiring and error-prone; simultaneously, it is a step with a great capacity for automation. Therefore, this work aims to develop an automatic system based on computer vision capable of diagnosing patients infected with VL through medical images. We compared the results obtained in this study with related works, where it was possible to observe that the methodology implemented here proved superior and more efficient, reaching an Accuracy of 99%. In this way, we demonstrated that the deep learning models, trained with images of the patient’s bone marrow’s biological material, can help specialists accurately and safely diagnose patients with VL.