Automatic segmentaion of multiple sclerosis disiese (MS) from magnetic rezonans images (MRI) for early detection

Multiple sclerosis, briefly referred as MS, is an autoimmune disease that holds on brain and spinal cord. As the reason is not known clearly yet, the immune system starts to defense the cells from themselves. The nerve cells (neurons) understands the myelin sheath (this can be defined as a kind of oily skin layer) around themselves as a foreign to the body and works for destroying it. This situation reveals a variety of nervous system symptoms. These symptoms are temporary, depending on the level of sickness, they can leave trace or disappear without leaving trace. MS is a clinical diagnosis. There have been no laboratory method for definitive diagnosis of the disease. Clinical characteristics of cases diagnosed is defined by how disease goes and by using laboratory methods. In some cases, clinical diagnosis could be easily done by examination and laboratory findings. Even in this case, other possible diseases should be exclude and the necessary laboratory investigations should be carried out. However, particular difficulties have been often experienced in patients who are in the early stage diagnosis. This period is important in the diagnosis of clinical and radiological follow-up commitment. Early diagnosis of MS is extremely important because with early detection the course of the disease could be changed and disease could be slowed down. The progression of MS means that the patient will start being disabled. The other importance of an early diagnosis for the patients who were diagnosed with MS is that they could have the chance of planning their own life. Magnetic resonance imaging (MRI) has been one of the most important method in the diagnosis of MS. MR imaging gives the opportunity of 3D inspection of the nervous system. White matter is shown very well. MR images is used for observing so-called MS plaques of various sizes in the white matter. MR is used for following disease. The purpose of this thesis is to develop a new application for determining multiple sclerosis lesions on brain T2 sequence magnetic resonance images by using segmentation methods. The reason of the choice of the magnetic resonance image as T2 images is that they are sequenced with high reflectance values for MS lesions in these images. In the thesis, in terms of having wider range of gray color scale 16-bit MR images were used. A total of 13 different MS patients were used in total 100 sections, 8 for testing the method and 92 for results. Application method consists of two stages. The first step is eliminating the skull bone tissue in T2-weighted brain images with the aid of SIOX algorithm, which has a very close reflectance values with MS lesions. The second step is detecting MS lesions which has the highest reflectance values from remaining brain tissue by the process of applying maximum entropy threshold. Implementation has the sensitivity value of 0.918, the positive predictive value of 0.837 and the accuracy value of 0.783 as calculated by the result analysis.