A comparison of image registration techniques in clinical dentistry

This work represents a comparative study for the methods of registration for dental panoramic X-ray images. The study tried to find the best registration technique used to register images of the jaw. Three image registration techniques are implemented and applied to dental panoramic X-ray images in both spatial domain and wavelet domain. The first technique is based on cross-correlation. The second one depends on selecting control points from both the reference and the input images. The last technique is based on maximization of mutual information between the two images. In the wavelet domain, discrete wavelet transform (DWT) has been performed to the reference image and the test image before the three techniques have been applied. Then the registration methods have been applied using approximation coefficients in the registration procedure. After finishing the registration steps, inverse discrete wavelet transform (IDWT) has been applied in order to obtain the registered image. The quality of the registration process was measured using the following criteria: normalized cross-correlation coefficient (NCCC) and percentage relative root mean square error (PRRMSE). Registration using maximization of mutual information (MI) has given the best results. It has the highest normalized cross-correlation coefficient (NCCC) and the lowest percentage relative root mean square error (PRRMSE). The NCCC reached 0.8513 and PRRMSE reached 5.0702e-4 % in the wavelet domain. Applying different registration techniques to dental panoramic X-ray images has proved that maximization of mutual information (MI) in the wavelet domain, is an effective tool for registration of X-ray images.