Classification of Malarial Parasite Using EfficientNetV2S Optimized with Humboldt Squid Optimization Algorithm
M. K. Vidhyalakshmi, S. Thaiyalnayaki, Allin Geo, G. Michael, G. Kalaiarasi · IETE Journal of Research · 2025
Malaria is a potentially deadly disease that is common throughout much of the world and is caused by bites from female Anopheles mosquitoes. The standard method of diagnosing malaria involves examining blood smears using a microscope to identify parasite-infected red blood cells. However, this typical approach is enhanced by using cutting-edge computer vision with deep learning algorithms. The main difficulty of the existing model lies in accurately classifying the malaria parasites from the blood smear data, which greatly affects the complete accuracy of the diagnosis. Therefore, this paper proposes a Classification of Malarial Parasite using EfficientNetV2S Optimized with Humboldt Squid Optimization Algorithm (CMP-EfficientNetV2S-HSOA). The input smear blood cell images are gathered from Malaria Cell Images dataset by National Institutes of Health (NIH). The input image is pre-processing by using Guided Edge-Aware Smoothing-Sharpening Filter (GEASSF) method to eliminate unwanted noises. The output of the pre-processed image is fed to the EfficientNetV2S for the classification of input smear blood cell images as parasitized or uninfected. Here, Humboldt Squid Optimization Algorithm (HSOA) is employed to improve the EfficientNetV2S classifier to provide accurate malaria classification. The proposed CMP-EfficientNetV2S-HSOA is implemented in Python platform. The proposed CMP-EfficientNetV2S-HSOA approach attains 23.31%, 18.62%, 35.26% higher accuracy and 19.36%, 30.37%, 22.53% higher specificity compared with the existing techniques: malaria parasite detection with deep learning approaches using CNN technique (MPD-DL-CNN), enhanced deep convolutional neural network of malarial parasite categorization (DCNN-MPC), and deep learning using random forest network in blood smear for categorization with malaria parasite diagnosis (DCNN-RF-CMP).