Related Experiment Video
Updated: Jun 16, 2026
09:53
In vivo Imaging of Transgenic Leishmania Parasites in a Live Host
Published on: July 27, 2010
16.0K
Assessment of Deep Learning Models for Cutaneous Leishmania Parasite Diagnosis Using Microscopic Images.
Ali Mansour Abdelmula1, Omid Mirzaei2,3, Emrah Güler4
1Department of Microbiology and Clinical Microbiology, Faculty of Medicine, Near East University, North Cyprus, Mersin 10, Lefkoşa 99010, Turkey.
Diagnostics (Basel, Switzerland)
|January 11, 2024
Summary
This study introduces a deep learning approach for diagnosing cutaneous leishmaniasis (CL). DenseNet-201 achieved high accuracy in identifying Leishmania parasites from skin smear images, offering a promising diagnostic tool.
Area of Science:
- Medical diagnostics
- Computational biology
- Parasitology
Background:
- Cutaneous leishmaniasis (CL) presents as skin lesions, increasingly prevalent in regions like Libya due to conflict and healthcare decline.
- Diagnosis of CL is crucial for effective treatment and public health management.
Purpose of the Study:
- To evaluate Convolutional Neural Networks (CNNs) for diagnosing cutaneous leishmaniasis amastigotes.
- To compare the performance of various pre-trained deep learning models for CL parasite detection.
Main Methods:
- Utilized a dataset of ultra-thin skin smear images from Leishmania-infected individuals.
- Employed pre-trained deep learning models: EfficientNetB0, DenseNet201, ResNet101, MobileNetv2, and Xception.
- Implemented a five-fold cross-validation strategy to assess model robustness.
Main Results:
- DenseNet-201 demonstrated superior performance among the evaluated models.
- DenseNet-201 achieved a mean accuracy of 0.9914.
- The model exhibited excellent sensitivity, specificity, and other key performance metrics.
Conclusions:
- DenseNet-201 is a highly effective deep learning model for diagnosing cutaneous leishmaniasis from skin smear images.
- This AI-driven approach offers a reliable and accurate alternative for CL diagnosis.
