This paper introduces a preliminary effort to develop an automatic image analysis method using artificial immune systems for clinical dental diagnosis. To diagnose dental deformity, especially malocclusion, manual measurement of certain geometry on the X-ray images is traditionally used, which relies on subjective judgment to determine the reference points. This paper proposes a feature extraction method that is based on the brightness distribution of the image instead of the anatomical parts. A negative selection algorithm is then applied to the data represented as real-valued vectors to detect the cases of severe malocclusion. Using the same data representation, one-class SVM was also tried to compare the detection capability with the negative selection algorithm. The results show that the negative selection algorithm appears more suitable for this problem.