Medical Image Segmentation Research Paper

Abstract— Medical image segmentation plays a vital role in one of the most challenging fields of engineering. It is indeed very helpful in the detection of the disease and its progressive treatment. More research has been enhanced for more effectiveness as far as the subject is considered. There are different methods used for medical image segmentation for medical images such as Thresholding method, Region Growing, Deformable Model, Clustering methods, Classifier, Markov Random Model etc. The main purpose of this paper is to provide a comprehensive reference source involved in Fuzzy C Means based medical image processing. There are different types of FCM algorithms and their advantages and disadvantages are discussed.

Key Terms: - Clustering;
In automated medical diagnostic systems, MRI (magnetic resonance imaging) gives better results than computed tomography (Cn as MRI provides greater contrast between different soft tissues in our human body. Therefore, MRI is much more effective in brain and cancer imaging [2].

Detection of brain tumor requires brain image segmentation Manual brain MR images segmentation is a difficult task. It requires plenty of time, non-repeatable task, non-Uniform Segmentation and also segmentation results may vary from expert to expert. So computer aided system is useful in this context. An automated brain tumor detection system should take less time and should classify the brain MR image as normal or tumorous accurately It should be consistent and should provide a system to radiologist which is self explanatory and easy to operate.

Automatic brain tumor detection and segmentation faces many challenges. It is indeed a difficult task to segment brain tumor in an automatic computerized system as it involves pathology and physics related
The images used for testing are of size 676x624 pixels, eight bits per color channel. Images that we have used for testing contain brain tumor of different size, shape and intensity. In order to check the accuracy of automated segmented tumor area, tumor from all images is segmented manually by the ophthalmologist. The manually segmented images are used as ground truth. The true positive rate is the ratio of number of true positives (pixels that actually belong to tumor) and total number of tumor pixels in the MR image. False positive rate is the ratio of false positives (pixels that don't belong to tumor) by total number of non tumor pixels in the MR image.

Figure 5 shows the experimental results for different MR images containing tumor of different shape and size. It shows that proposed method have extracted the brain tumor accurately.

The results of tumor segmentation for MR images are summarized in table-I. It shows the results in terms of average accuracy and their standard deviation as compared with ground truth. Average accuracy is computed by counting the total number of pixels correctly classified. IV.