One of the important factors of medical diagnosis is using safe and non-invasive method. Infrared thermal imaging, as a non-invasive method, attracted scientists' attention over the past decades and it's used in many medical diagnosis. This method of imaging can provide useful information from patient's body. Some of the imaging methods which is currently used are invasive and often expensive. One of the important applications of infrared thermography is in Rheumatology field for diagnosis and assessment of inflammatory joint diseases. Inflammatory joint diseases, which contain Rhuematoid Arthrits, are very common worldwide nowadays and lots of people are dealing with them. Some of these disease are auto-immune. In the other words, the immune system of the body attacks the joint for an unknown reason. The cause of these kind of diseases are unknown yet and consequently developing a method for early diagnosis of Rheumatoid Arthritis can be very important. Usually this disease doesn't have any special symptoms in the early stage and this makes its diagnosis and assessment difficult. It is notable that this disease doesn't have any special treatment and can only be stopped by special drugs. In the last stages, the bones which are connected to the joints are completely deformed causing a great disability for the patient. Normally the heat distribution over the skin is symmetric and any asymmetric pattern can be a sign of special disease. In this thesis a method based on the mutual information is developed for analyzing symmetric pattern in the thermal images taken from patients. In the proposed method, images are registered first. The joint histogram of image pair is estimated using Non-parametric windows and mutual information is then calculated from the estimated joint histogram. To improve the efficiency of the proposed method, mean temperature as the second feature is defined. Finally ltr" Keywords: 1. Rheumatoid Arthritis, 2. Infrared thermography, 3. Non-parametric windows, 4. Mutual information
