A machine learning-based system for detecting leishmaniasis in microscopic images

Area of Science:

• Computational biology
• Parasitology
• Medical diagnostics

Background:

• Leishmaniasis is a globally significant parasitic disease, second only to malaria in mortality.
• Current diagnostic methods rely on microscopy, which is time-consuming and prone to errors.
• There is a critical need for more efficient and accurate diagnostic tools for leishmaniasis.

Purpose of the Study:

• To develop an artificial intelligence (AI)-based algorithm for the automatic diagnosis of leishmaniasis.
• To create a system capable of detecting leishmania parasites in images.
• To provide a faster, more accurate, and cost-effective diagnostic solution.

Main Methods:

• The Viola-Jones algorithm was adapted for leishmania parasite detection.
• Key steps included Haar-like feature extraction and integral image creation for efficiency.
• The adaBoost technique was employed for feature selection and classifier training.

Main Results:

• The AI system achieved 65% recall and 50% precision in detecting leishmania parasites within macrophages.
• For amastigotes outside macrophages, the system demonstrated 52% recall and 71% precision.
• The algorithm shows promise in identifying parasitic presence in diagnostic images.

Conclusions:

• The developed AI system is accurate, fast, user-friendly, and cost-effective.
• Artificial intelligence presents a viable alternative to conventional leishmaniasis diagnostic methods.
• AI-powered tools can significantly improve the efficiency and reliability of leishmaniasis diagnosis.