EXPLORE PUBLICATIONS BY COUNTRIES

Malaria, Histogram of Oriented Gradients, Principal Component Analysis, Support Vector Machine.

Malaria is an infectious disease caused by Plasmodium parasites, with the P. falciparum species being the primary cause of mortality globally. Conventional microscopy-based approaches for malaria diagnosis possess considerable drawbacks, such as prolonged analysis durations and a requisite high degree of expertise. This study proposes an artificial intelligence-based methodology that integrates feature extraction via Histogram of Oriented Gradients (HOG) and dimensionality reduction through Principal Component Analysis (PCA) to enhance malaria detection efficacy utilizing a Support Vector Machine (SVM) model. This research additionally contrasts the efficacy of this approach with that of the Convolutional Neural Network (CNN) method. This study utilizes two dataset sizes, comprising 200 and 2000 photographs, with 80% allocated for training and 20% for testing. The experimental findings indicate that the fundamental SVM model attains 67.5% accuracy on short datasets, which increases to 90% with HOG, but decreases to 80% with PCA. In extensive datasets, the fundamental SVM model attained an accuracy of 70.5%, which increased to 87% using HOG, but subsequently decreased to 81.25% following PCA. Compared to the CNN method, which achieved 97% accuracy, it exhibited superior generalization capability on the test data. This work illustrates that the integration of HOG and PCA enhances malaria detection efficacy, albeit with certain trade-offs. This work examines whether hybrid classical methods, such as HOG and PCA in conjunction with SVM, may provide efficient and accurate malaria detection, particularly in resource-limited settings, emphasizing its practical applicability in real-world scenarios.