Segmentation and Feature Extraction for Malaria Detection in Blood Smears

Abstract

Malaria remains a critical global health challenge, particularly in tropical and subtropical regions. Early and accurate diagnosis is essential for effective treatment and control. Traditional methods of malaria diagnosis, such as microscopic examination of blood smears, are time-consuming and prone to human error. This study aims to develop an automated system for malaria detection using machine learning techniques, specifically a decision tree classifier. The dataset, sourced from the National Institutes of Health (NIH), comprises 27,558 blood smear images equally divided into Normal and Malaria classes. The preprocessing steps included segmentation using the Canny edge detector and feature extraction using Hu Moments, followed by data normalization to ensure a mean of 0 and variance of 1. The decision tree classifier was trained and evaluated using 5-fold cross-validation, yielding an average accuracy of 77.32%, precision of 77.31%, recall of 77.37%, and F1-Score of 77.48%. These results demonstrate the model's robustness and effectiveness in differentiating between malaria-infected and uninfected images. The study confirms the viability of using Hu Moments for feature extraction and highlights the decision tree classifier's suitability for this task. The proposed method has significant implications for automated malaria diagnosis, potentially improving diagnostic accuracy and efficiency in clinical settings. Future research should validate these findings on diverse datasets, explore advanced classification techniques, and integrate real-time image acquisition to enhance practical applicability. The integration of such automated systems in healthcare can revolutionize malaria diagnosis, especially in resource-limited settings.