Thyroid Nuclide Image Segmentation Combined With Super-Reconstruction And KFCM

In recent years, the growth rate of domestic sharp rise in thyroid disease, thyroid cancerranked first in the rate of growth of various types of cancer, and has great harm to humanhealth. Early thyroid cancer is generally expressed as thyroid nodules, such as the timelyidentification of nodules from benign thyroid scintigraphy in malignant, can be completelycured after treatment, so the image is an important basis of radionuclide diagnosis andtreatment of thyroid disease.A large number of images produced by the thyroid scintigraphy makes it difficult todetect thyroid nodules manual, at present most doctors determine the edge of the area ofinterest through manual segmentation method, manually segmented edge imprecise,subjectivity, less efficient. Visible radionuclide image segmentation directly affects the targetexpression and subsequent measurement and analysis of thyroid nodules.Radionuclide blurred images with uneven tissue, poor spatial resolution, gray lowcontrast, low SNR characteristics, in order to obtain a clear, smooth segmentation boundaries,for the direct target edge after the division is rough and the existence of jaggies, thesuper-resolution reconstruction algorithm based on sparse representation is adopted, toimprove the spatial resolution of the image, making the details of the radionuclide imageorganization richer. For high quality reconstructed image segmentation nuclides, theexperimental validation of traditional thyroid radionuclide image segmentation method is notapplicable. Therefore, the algorithm suitable to solve the problem of the existence ofambiguity and uncertainty gray fuzzy clustering image segmentation is adpoted, the standardFCM algorithm considers only the pixel values from the cluster center, which does notconsider the impact and use of space adjacent image pixels information, meanwhile kernelimage feature has good adaptability and immunity to noise, the fuzzy clustering algorithm forthyroid radionuclide image segmentation is adopted, and the comparison with the FCMalgorithm on a number of iterations and time consuming is made.As can be seen from the experimental results, the fuzzy clustering algorithm has feweriteration and shorter running time s than FCM, meanwhile a smoother edge region of interestis detected. Thus effectively assist doctors to improve the detection accuracy and efficiency ofthyroid nodules, reducing the rate of missed diagnosis and misdiagnosis.