Database and Images of Cutaneous Leishmaniasis Direct Smears: Detection of Leishmania spp. Parasite in Microscopy Images Using Machine Learning Algorithms.

This database and images were created for Cutaneous Leishmaniasis research, mainly for computer vision and machine learning studies. The dataset is organized into two main folders, designed to facilitate access and analysis of the images and corresponding labeling data. The first folder, 'EPVIEW_LC_S,' contains a total of 500 PNG images, each with a resolution of 2592 x 1944 pixels. These images represent positive direct smears with Cutaneous Leishmaniasis. The organization of the images follows a hierarchical structure, distributed across 50 folders (L1 – L50), where each folder symbolizes an individual exam slide. Inside each main folder, there are two subfolders that house the images of the two samples present on each slide, with 5 images per subfolder, corresponding to the 5 microscopic fields captured for each sample. This results in a set of 10 images per slide. The second folder, 'Etiquetado_LC_T,' contains the labeling process carried out for each image with the collaboration of two experts in the field. A total of 7,905 Leishmania spp. parasites were identified and labeled. Using bounding boxes, the precise dimensions of the parasites were recorded, including width, height, and the central location of each parasite within the image, with (x, y) coordinates, along with the parasite label. This information was stored in JSON files, whose names correspond directly to the images they reference, ensuring an accurate relationship that facilitates the use of the data for research purposes. The organization structure of the labeling folder mirrors that of the image folder.