Comparative performance analysis for machine learning techniques on malaria detection using CNN and microscopy images
T. Priyanka Kumari, Dakkili Maneiah, V. Somalaxmi, Kothapally Subramayam Chari, R. S. Sabeenian · 2025
Parasites that cause malaria enter the body when an infected mosquito bites a human and produce the potentially deadly disease. If automation diagnosis this condition, accurate diagnosis can be made, and therefore reliable health care can be provided in resource-poor locations. Currently, the diagnosis of malaria is based on examination of patient blood smears under a microscope. In addition, it is time-consuming and the diagnosis relies on analytical expertise and experience of the examinations. Image processing-based automatic image recognition systems have previously been used for malaria blood smear diagnosis. This study provides the best methods for distinguishing between infected and no infected cells both accurately and computationally; Several ML models such as CNN, SVM, LR, RF, and KNN were evaluated on the labeled dataset of microscopic images of blood smears. Performance parameters such as computational time, accuracy, precision, recall and Fl-score are used to compare these algorithms. According to the results, CNN based deep learning models achieved the highest accuracy of 97.8%.