Automatic detection of Visceral Leishmaniasis in humans using Deep Learning

Leishmaniasis is a commonly neglected disease present in tropical and subtropical countries, affecting 1 billion people. Visceral Leishmaniasis (VL) is the most severe form and can lead to death if left untreated. In this work, we apply deep learning techniques to detect VL in humans through images of slides from the parasitological examination (microscopy) of the bone marrow, aiding in an automatic and accurate diagnosis. This work investigates five deep learning architectures combined with preprocessing, data augmentation, and fine-tuning techniques to detect this disease in images. We compared our results with five related state-of-the-art works, which showed that the proposed classification method surpassed them in all metrics. We achieve an Accuracy of 98.7%, an F1-Score of 98.7%, and a Kappa of 98.7%. Therefore, we demonstrated that trained deep learning models with microscopic slide imaging of bone marrow biological material could precisely help the specialist detect VL in humans.