Registration Of SPECT Image And B-type Ultrasound Image Based On Genetic Idea Bat Algorithm

With the increase of incidence of thyroid carcinoma, correct detected the benign or malignant of thyroid nodule is of great significance. Because only about five percent of the thyroid nodule is malignant and most thyroid nodules are benign, timely detection and treatment can be cured completely usually. B-type ultrasound can clearly show that subtle changes, it is no-radiation, accurate positioning, but not sure of benign and malignant thyroid nodules. SPECT imaging is functional imaging, can according to the distribution of iodine-1311 in the glands, providing of benign and malignant lesions and the function of the close relationship between the images, but to the positioning of the nodules are inaccurate. In order to provide complementary information of the thyroid, proposed to thyroid SPECT image and B-type ultrasound image registration and fusion. This paper mainly studies the registration of the thyroid SPECT image and B-type ultrasound image, in preparation for fusion of the two kinds of pattern images.SPECT image and B-type ultrasound image have quite different gray level, which means the registration method based on gray level is inapplicable. In view of the two images contain the outline characteristics of the thyroid and tumor, this paper take the registration method of combining gray and characteristics. We use the KFCM method as the segmentation of SPECT image, the GCBAC and NSST method is proposed as the segmentation of B-type ultrasound image.And then the binary image of outline of the thyroid and tumor to prepare for the thyroid image registration.For gradient normalized mutual information having more local extreme in far away from registration place, we improve the gradient normalized mutual information in which the part is not change with the location of the molecules into the mutual information. In view of the bat algorithm has low convergence accuracy and easy to fall into local extreme, this paper put forward genetic idea bat algorithm which combine the selection, crossover and mutation operation of genetic idea and the bat algorithm. Improved gradient normalized mutual information as the similarity measure and the genetic idea bat algorithm is used to optimizethe space transform parameters, the experimental results show that the method has high locating accuracy and rate of registration of SPECT image and B-type ultrasound image.