| At present, thyroid nodule has a high incidence, and some of the thyroid nodules belongto thyroid malignant tumor. The main means of treatment of thyroid cancer is surgicalremoval, in order to get the best effect of surgery the doctors must diagnosis of benign andmalignant tumor of thyroid in time. B-ultrasonic examination and nuclide scan are two kindsof medical imaging technology in diagnosis of thyroid tumor, and the two methods have theirown advantages and limitations. Thyroid B-ultrasound images can get the thyroid anatomicshows and focus on express morphological change. Nuclide images can provide the change ofthe function. The doctors can diagnose the pattern of the nodule by nuclide images. Fusing thetwo kinds of image can complement each other and help doctors diagnose the thyroid nodulecomprehensively.The image fusion method based on multi-scale analysis is the most popular way offusion at present. Wavelet transform, Contourlet transform and NSCT transform are typicalmulti-scale transform. Using the three transforms to fuse the B-ultrasonic image and nuclideimage of thyroid and results show that the wavelet transform is more suitable for image fusionof thyroid. In the process of the experiment, the choices of the wavelet decomposition leveland the wavelet base have been discussed.On the basis of image fusion based on wavelet transform, introducing artificial fishalgorithm in image fusion. Fusing high frequency component with the fusion rule ofmaximum absolute value after perform wavelet transform on the original images, and usingweighted fusion algorithm to fuse the coefficient of low frequency part, and the optimalweights is obtained by using artificial fish-swarm algorithm. In order to make the fusionresults contain enough information, we choose the associated entropy as the adaptive value inthe process of artificial fish-swarm algorithm optimization. Experiment proved that themethod has improved the quality of the fused image. |
PDF Full Text Request