Identification of Plasmodium Vivax in Blood Smear Images Using Otsu Thresholding Algorithm

In this research, we explore the efficacy of Convolutional Neural Networks (CNN) and Support Vector Machines (SVM) in identifying Plasmodium vivax from blood smear images. We utilized a dataset comprising images of Plasmodium vivax and non-infected cells, applying CNN for deep feature extraction and SVM with otsu’s thresholding for segmentation. The dataset was preprocessed and augmented to enhance model performance. The CNN architecture, consisting of multiple convolutional and dense layers, achieved an accuracy of 98.46% on the validation set. For comparison, features extracted using Otsu’s Thresholding were fed into an SVM classifier, yielding an accuracy of 82%. Confusion matrix was generated to evaluate the classification performance of both models. The CNN model demonstrated superior accuracy and robustness in classification tasks compared to the SVM model. This research shows how deep learning frameworks can be used to analyse medical images and how important it is to have methods for extracting and choosing features to make machine learning models work better.