Manuscript received March 10, 2025; revised May 12, 2025; accepted June 9, 2025; published July 17, 2025.

Abstract— Malaria has been recorded as one of the deadliest diseases globally. Accurate diagnosis is essential for suitable treatment, and the traditional practice of malarial diagnosis has proved inefficient as results depend on the skills of the health personnel. Deep learning models have recently proven helpful in the rapid detection of malaria parasites. This research focused on developing a classification model of Convolutional Neural Networks (CNN) architectures and comparing these models to identify the most effective one for automatic malaria parasite detection on thin blood smear images. A dataset of 27,558 digital blood images was collected from the National Institutes of Health (NIH) database in Bangkok, Thailand. The dataset was categorized into parasitized and uninfected cells and was fragmented into training (80%) and validation (20%) sets. Performance metrics for measuring the model’s performance include sensitivity, specificity, precision, and F1−Score. The model predicted and classified thin blood smear digital images as either parasitized or uninfected with custom InceptionV3 outperforming the VGG19 and custom CNN with an accuracy of 89.85%. The result shows that malaria diagnosis on microscopic thin blood images using deep learning can potentially improve early detection of malaria parasites, which could prevent deaths, reduce the workload of Parasitologists, and eliminate other limitations of the traditional malaria diagnostic approaches.

Keywords—convolutional neural network, malarial parasite detection, classification model, Diagnosis, Digital blood images

Cite: Matthew C. Okoronkwo, Chikodili H. Ugwuishiwu, Boniface Emmanuel, Collins N. Udanor, Charles Ikerionwu, Osondu E. Oguike, Nnaemeka E. Ogbene, Rita N. Nweke, Folake O. Adegoke, Kenneth Ugwu, Ignatius I. Ayogu, and Anthony C. Ike, "A Comparative Study of Convolutional Neural Networks (CNN) Architectures on Microscopic Blood Film Images for Malaria Diagnosis," Journal of Image and Graphics, Vol. 13, No. 4, pp. 348-361, 2025.

Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC-BY-4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.