Research In Medical Diagnosis Of Malaria Based On A New Faster Unet Attention Model

Malaria is one of the deadliest parasites in the world and across the African continent.Despite the measures to limit this parasite,according to data from the World Health Organization(WHO),the world has observed the evolution of 2million infections between 2020 and 2021.The detection of this parasite is a task really tedious for researchers in areas with high contamination,the traditional method of detecting this parasite requires a lot of time and resources.This analysis requires the use of blood samples from potentially infected patients to be observed under a microscope.The image of the result obtained makes it possible to distinguish the parasites.This time of realization is explained by the lack of skilled labor on the one hand,and on the other hand the lack of adequate tools for the realization of this task.In order to provide a faster measure to detect this parasite in blood sample cells,a new segmentation method based on the use of a faster Unet attention model was proposed.This method allows more accurate aggregation of contextual information by introducing the attention-augmented convolution mechanism.This mechanism significantly improves semantic segmentation tasks on the difficult segmentation of images of blood smears infected with malaria parasites.The performance experiment shows that our model provides better results than a basic VGG,Res Net,Inception V3,or Unet model(nearly 94%)despite its limited number of data.