Brain Tumor Detection using Tree-based Representation of Fuzzy Sets in MR Images

Nowadays image processing is a common, useful and effective tool for physicians, to examine inner organs of human body. Decision behalf functional medical images needs expert operators to catch high quality images time and also expert radiologist to detect and report critical point of images. In this paper we want to use fuzzy connectivity concept to perform on different images to get a fuzzy connectivity map. These fuzzy connectivity map would help us detecting tumor regions in the images. We proposed a new scanning approach for connectivity measurement in fuzzy sets of image, which significantly decreases the copmutation time. Applying inclusion filtering with a predefined marker led to seperation of a specific leave of the tree. An automatic procedure for determining such markers is proposed based on Hough transform with regard to specific geometrical shape of brain tumors. Therefore the leave which is related to the tumor is detected and the corresponding region is segmented in the image using defuzzification process. The results show that the acuuray of the proposed method is more than two conventional algorithms, thresholding and Hough transform. Although error rate increases in comparison with region growing our proposed method is fully automatic. The result of adding noise to the images also show an acceptable robustness of the proposed method

Brain Tumor Detection using Tree-based Representation of Fuzzy Sets in MR Images Keywords:

Fuzzy sets , connected components , Tree-based representation , Defuzzification , MRI images , Tumor detection