Identifying abnormalities in Computed Tomography brain images using symmetrical features
W M. Diyana, Wan Mimi Diyana Wan Zaki, CunRui Kong · 2009
This paper proposes an automated method to identify abnormalities by exploiting symmetrical property features in Computed Tomography (CT) brain images. This method consists of two main steps; symmetrical axis detection and rule based abnormalities detection. Based on the principle axis theorem, any tilted intracranial is firstly corrected before symmetrical axis is generated. Then, segmented CT brains intracranial are divided into left half and right half used to produce possible feature vectors. Size (area) and location (centroid) of the abnormalities are chosen as main features for the development of the rule based abnormalities detection system. This experimental work uses twenty abnormal and eighty normal CT brain images and performance of proposed method is evaluated in term of sensitivity and specificity. It shows that the proposed automated method using symmetrical features proved to be efficient and accurate, and gives reliable results for every CT brain image tested.