A HYBRID APPROACH FOR MALARIA CLASSIFICATION USING CNN-BASED FEATURE EXTRACTION AND TRADITIONAL MACHINE LEARNING CLASSIFIERS

A HYBRID APPROACH FOR MALARIA CLASSIFICATION USING CNN-BASED FEATURE EXTRACTION AND TRADITIONAL MACHINE LEARNING CLASSIFIERS

Malaria is a major global health threat, and timely and correct diagnosis is essential for effective treatment. Traditional diagnostic methods, such as the microscopic examination of blood smears, are time-consuming and require expert personnel. The study presents a mix of machine learning methods...

Full description

Saved in:
Bibliographic Details
Main Authors: omar Mohammed Alzakholi, Walat A. Ahmed, Bafreen N. Mohammed, Asaad Kh. Ibrahim
Format: Article
Language:English
Published: University of Zakho 2025-07-01
Series:Science Journal of University of Zakho
Subjects:
Malaria classification
Convolutional Neural Networks (CNNs)
Web vulnerabilities, feature extraction, feature selection, data mining, supervised machine learning, phishing websites, cyber security, and phishing identification
Malaria classification, Convolutional Neural Networks (CNN), Feature extraction, Random Forest, Logistic Regression, KNN, SVM, Gradient Boosting.
Heart Disease Prediction, Logistic Regression, Support Vector Machine, Feature Selection, Linear Discriminant Analysis, Cleveland data.
KNN estimator
Online Access:http://sjuoz.uoz.edu.krd/index.php/sjuoz/article/view/1489
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Malaria is a major global health threat, and timely and correct diagnosis is essential for effective treatment. Traditional diagnostic methods, such as the microscopic examination of blood smears, are time-consuming and require expert personnel. The study presents a mix of machine learning methods for automatic diagnosis of malaria by using the feature extraction capability of Convolutional Neural Networks (CNNs) along with the efficient classification performance of traditional machine learning classifiers. For our study, we utilize VGG16 CNN with a weight pre-trained on ImageNet to extract the features from non-infected and infected blood cell images from malaria. Five classical machine learning algorithms, such as Random Forest, Logistic Regression, K-Nearest Neighbours (KNN), Support Vector Machine (SVM), & Gradient Boosting, are used to classify the extracted features. Each classifier's performance is calculated based on accuracy, F1 score, precision, and recall metrics. The results of our experiments showed that the hybrid model has high accuracy in classification, where the Logistic Regression classifier could achieve above 93% accuracy. This hybrid method is a powerful diagnostic for malaria disease, accomplishing a more satisfactory compromise between the efficacy of the deep learning architectures such as CNNs, and the computational capabilities of more conventional classifiers. It holds promise for deployment into resource-limited settings where fast, automated threading diagnostic systems are much needed
ISSN:2663-628X
2663-6298

Similar Items