Detection of Malaria Parasites in Thick Blood Smear Images using Shallow Neural Networks and Digital Image Processing Techniques

In routine malaria microscopic diagnosis, the thick blood film exam is performed to detect malaria parasites. The microscopic diagnosis is based on counting the number of parasites (parasitemia) in stained blood slide. The most recent work published for malaria parasite detection and counting points in the direction of using complex neural network architectures, such as Efficient NET or ROENet convolutional networks, or specialized networks for object detection, such as YOLO and Faster- RCNN. Such models make the methods difficult to use on mobile devices. Simple models could run in mobile devices, mainly in regions of low-and-middle-income country, with poor internet access This work aims to present simple and fast models for malaria parasite counting and detection, based on shallow neural networks, as one layer perceptron, logistic regressor and multilayer perceptron, associated with digital image processing technique. The best results obtained, an F1-score of 97.57%, is better than other results previously published in the literature.