Enhanced Image Analysis for Detecting Malaria Infection in Blood Samples

The methods of detecting malaria infected cells from microscopy images are discussed in detail in this paper. Malaria disease being a world health problem today, there is need to develop better diagnosing technique because of the high death rate that has been plunged to this disease. Having used a dataset of 27,558 images as infected and uninfected, this study employs Convolutional Neural Networks(CNN), Fuzzy and VGG19 models for semiautomated classification with higher speeds and accuracies as opposed to other methods. It consists of mandatory phases, which predetermine preparing the data set for training, including resizing, normalization, and data augmentation. Among to feature selection, SelectKBest method is used in order to enhance input features of the model. Initial experiments with the CNN model again revealed high accuracy greater than 90%, thereby confirming the use of the model in real–life diagnosis of malaria. In addition, the study emphasises the importance of feature extraction and the use of machine learning in the diagnosis function. The outcomes support the adoption and use of automated detection techniques as the human factor contributes greatly to errors in diagnosis, practices should embrace such technologies leading to the fight against malaria.