This repository contains the source code and data utilized in the development of A deep learning-based model for detecting Leishmania amastigotes in microscopic slides: a new approach to telemedicine.
In this research, LeishFuNet, a deep learning model that employs same-domain transfer learning techniques with various pre-trained deep models, was developed. LeishFuNet is an explainable model, as GradCAM was applied to its predictions, demonstrating that the model analyzes input images in a manner similar to clinicians. This explainability is crucial, as the model's trustworthiness is vital in healthcare-related studies. Another significant contribution of this research is the publication of a novel dataset of Leishmania microscopic images. Given the scarcity of datasets in this specific field, this newly published dataset can be highly beneficial for future studies.
Dataset This directory contains the microscopic images used in this study.
fusion_utility This directory includes functions utilized in leishfunet directory to develop the main model of this research.
gradcam This directory contains the code and results of implementing the GradCAM technique on LeishFuNet's predictions.
leishfunet This directory contains the code used to develop the main model for this study.
medical-pretrained This directory contains the code used to develop the pretrained medical models for further same-domain transfer learning implementation in the leishfunet directory.
mobilevit This directory contains the code for implementing an advanced deep learning model on the dataset used in this study.
utility This directory includes functions utilized in medical-pretrained directory to develop the pretrained medical models.
If you utilize the models or dataset from this research, please cite:
Sadeghi, A., Sadeghi, M., Fakhar, M. et al. A deep learning-based model for detecting Leishmania amastigotes in microscopic slides: a new approach to telemedicine. BMC Infect Dis 24, 551 (2024). https://doi.org/10.1186/s12879-024-09428-4