Artificial Intelligence in Microscopic Diagnosis of Malaria: A Laboratory Study in Pakistan

Abstract

Background: Malaria remains a major burden in developing countries including Pakistan; early and correct diagnosis is crucial for the treatment and control of the disease. The standard method of malaria diagnosis, microscopic examination of Giemsa-stained blood smears, is highly dependent on the skill of the microscopist. Objective: A recent study set out to design and test a malaria diagnostic tool using artificial intelligence (AI) to detect malaria parasites in peripheral blood smears in Pakistan. Method: A cross-sectional study was performed in the laboratory on blood smears from patients with suspected malaria. Thick and thin blood smears, prepared according to protocol were stained with Giemsa and scored on light microscopy (the gold standard). Images were taken of the stained smears and the result from the microscopy was used to label the image. The images were used to train and test a convolutional neural network (CNN) artificial intelligence (AI) model for malaria. We determined the diagnostic performance measures, accuracy, sensitivity, specificity, precision, F1-score and area under the receiver operating characteristic curve (AUC). Result: The AI model showed strong agreement with conventional microscopy and demonstrated promising diagnostic performance in identifying both malaria-positive and malaria-negative samples. The model exhibited high accuracy, sensitivity, and specificity, supporting its potential as a diagnostic tool. Conclusion: This research shows that AI-assisted microscopy could be used as an adjunct technique to improve the speed, accuracy and efficiency of malaria diagnosis, particularly in resource poor settings. AI-assisted microscopy could be used as an adjunct to existing methods in Pakistan but first we need to test it on a larger scale.

Keywords: Malaria, Artificial Intelligence, Microscopy, Deep Learning, Pakistan, Blood Smear Diagnosis, Plasmodium falciparum, Plasmodium vivax