Download PDFOpen PDF in browserMalaria Detection Using ANNEasyChair Preprint 44185 pages•Date: October 17, 2020AbstractMalaria is a life-threatening disease so rapid, accurate diagnosis is required to control the disease. The detection of Malaria parasites is done by pathologists manually using microscopes. The manual microscopic examination is still the gold standard for malaria diagnosis, it is tedious, time consuming and requires special training and considerable expertise, where the lack of access to malaria diagnosis is largely due to shortage of expertise. The identification and counting the number of RBCs in the image is very important to diagnose. In this study proposed a system to detects and identification malaria parasites and counting the number of RBCs in the image from microscopic images. The dataset was collected from CDC dataset site. The proposed system began by pre-processing, to remove unwanted objects and noise from the image by morphological reconstruction. After pre-processing the image region of interest (Erythrocyte or RBCs) was segmented from the background by using Marker controlled watershed method applied on green channel color image. Next, the cell which has a plasmodium was extracted by k mean cluster. Then the statistical features were used in all of the cells to show the infected cells, and healthy cells. The ANN classified the normal and infected cells. The framework was tested to classify the RBCs in blood cell images, and the results are; accuracy (99.3%), sensitivity (98.6%), and specificity (94.8%). Keyphrases: Back Propagation, K-means, Segmentation, morphological reconstruction, parasite
|