Study On Automatic Segmentation Of Inner Ear And Its Application Based On MRH

The main part displayed in MRH is lymph which is inside and outside of the membranous labyrinth, which is complementary to bony labyrinth displayed in CT.The segmentation of the inner ear based on MRH is a preoperative part of the CI, and a necessary method of quantitative measurement on the inner ear structure. Automatic segmentation of the inner ear has some problems:the grayscale intensity of the inner ear is similar to the neighbor cerebrospinal fluid; The inner ear has small, complex and intricate structure; Automatic location of it is difficult. Analyzing and quantifying the inner ear, however, is an extremely laborious and time consuming task if done manually. Automatic segmentation based on digital image processing computer-aided technology is of great significance for improving the quality and efficiency of the inner ear segmentation. To settle the problems above, the automatic segmentation of the inner ear and its application studied in this thesis and the main contents are as follows:The ROI of the inner ear is proposed to be extracted using the registration method, and the location accuracy is about2voxels. Combined with the tubular structural features, the inner ear is enhanced and the background is suppressed by the enhancement factor defined by eigenvalues of the three-dimensional Hessian matrix. Then, the enhanced result segmentation uses the fuzzy cluster method. Furthermore, the pre-segmentation result is optimized using a hybrid segmentation method in level set formulation. The voxels overlap rate of the final results with the doctor hand-segmentation is high. Comparing to the manual segmentation which takes approximately ten minuses for a single inner ear, the automatic approach on average takes less than three minutes to complete, and gives more consistent results across datasets.Based on the segmentation results, the centerlines of the inner ear are extracted using the fast marching method. By calculating length from the round window to the top of the cochlea, the artificial cochlear electrode depth is estimated preoperatively. It can be used for the electrode selection and the guidance for the surgery; Some parameters of semicircular canals are calculated and provide evidence for the related diseases.