Pruned Associative Classification Technique for the Medical Image Diagnosis System

Brain tumor is one of the leading cause of death in recent years. This paper proposes the tumor detection in CT scan brain images, which can assist the medical image diagnosis system. The method proposed here makes use of association rule mining technique to classify the CT scan brain images. It combines the low-level features extracted from images and high level knowledge from specialists. The proposed system consists of: a pre-processing phase, feature extraction phase, a phase for mining the resultant transaction database, a final phase to build the classifier and generating the suggestion of diagnosis. The classifier built in this method has an important characteristic that it can suggest multiple keywords per image, which improves the accuracy. Experimental results on pre-diagnosed database of brain images shows high accuracy (up to 95%), allowing us to claim that the use of associative classifier is an efficient technique to assist in the diagnosing task.