ANN Algorithm for Brain Hemorrhage Detection Using CT Images

Abstract

Brain hemorrhage is a type of bleeding that might occur in more than one location in the brain. It is usually imaged by Computed Tomography (CT). These images of hemorrhage allow accurate disease prediction and efficient patient assessment of morphological changes in the brain as the recovery progresses. CT image processing is commonly used to enhance the visual details on an image. This study examines the ability of Artificial Neural Network (ANNs) classifier to detect the existence of brain hemorrhage based on CT images. The proposed algorithm was tested on 200 brain CT images; the dataset contained 100 normal and 100 abnormal cases. The results indicate a sensitivity of 79.8%, a specificity of 82.3%, and a classification accuracy of 81.0%. This algorithm can be used as a guiding tool for trainee radiologists to test expert diagnosis and to minimize mistakes in the current techniques. In conclusion, the classifier is useful to classify the images, and helps to diagnose the disease automatically without manual guidance by the users.