Research On Thickness Measurement Of Femoral Head Cartilage Based On Mri

As one of important parts in human hip joint, the femoral head has high probability to get lesion or even to become necrotic. This is still one of the knotty problems in medical field at present. The damage on the femoral cartilage is one of the major features in early stage of diseases on the femoral head. Thus, it is very important to evaluate the articular cartilage timely and accurately in order to make a diagnosis and to determine the therapeutic method. With the advantage of no damage of ionizing radiation, and being sensitive to the cartilage tissue, the technology of Magnetic Resonance Imaging is regarded as the best noninvasive method of cartilage detection and evaluation at present. Using the computer to analyze and process the MR images, and to estimate and evaluate the thickness of cartilage, has been gradually considered as a research hotspot of medical image processing. Research on this technology will greatly improve the efficiency of diagnosis and the effect of treatment, which is also coincident with the development trend of medical image processing. In this thesis, the main research content is several key detection techniques in MR images for femoral cartilage, including image segmentation, thickness measurement, 3D reconstruction.On the aspects of image segmentation, combining respective advantages of various methods, a multi-step segmentation based on information fusion was proposed to implement fast automatic segmentation of the femoral cartilage in MR images. After pretreatment, the MR images were used to calculate the center of the femoral head by imposing the Hough transform. Combining the anatomic constrain of the femoral head, the region of interest (ROI) was selected and the rough segmentation was realized. The image edges of the object region were then extracted using the adaptive thresholding Canny detector. According to the properties of the pixel on femoral cartilage edge, we labeled these edges and removed the noise edges according to the custom rules to acquire the exact edges of the femoral cartilage. Finally, the femoral cartilage was segmented by extracting the image information between the cartilage edges.On the aspects of thickness measurement, the main technique at the present time is to measure the thickness of cartilage in the MR image plane. By building mathematical modal of 2D sheet structure and simulating signals in the process of the MR photographing, it was proved theoretically that the zero-crossing method based on the second directional derivatives of Gaussian blurring was effective to measure the sheet structure thickness in 2D MR images. The simulation experiments and MR photographing experiments further verified the conclusion. Based on it, aiming at the problem of overestimation due to the surface shape of cartilage and the slice imaging, a correcting algorithm based on shape constraint was proposed to correct the in-plane thickness, and the real thickness value of femoral cartilage in the 2D MR images was obtained accurately.Measuring the cartilage thickness of femoral head directly from the 3D MR images will be the trend in the development of measuring technology. Without need of correction, this method is more significance physically in the principle of measurement. The distance between neighbor slices in MR images is much bigger than that between the neighbor pixels in the slice image, so it's necessary to interpolate between the image slices in order to performance the spatial measurement. A mixed interpolation algorithm based on segmentation of gray region was proposed, which combined the efficiency of linear interpolation and the accuracy of matching interpolation organically, and achieved a good interpolation of the images among the slices to make the voxel isotropic. On the basis of this, the zero-crossing method based on the second directional derivatives of Gaussian blurring was extended to 3D space. The complex 3D convolution of Gaussian function was simplified by Hessian matrix to realize the measuring the spatial thickness of cartilage directly. Experiments showed that this is the affective method of measurement.On the aspects of 3D reconstruction, the classical triangulation methods were used to reconstruct the surfaces of the femoral head cartilage. Aiming at the problems of surface roughening caused by the dense quantitative points data set, a data points selection method based on cross dislocation was proposed to eliminate the terrace effect on the reconstruct surface successfully. To solve the problem of time consuming during the spatial direct triangulation, an improving Delaunay triangulation method was proposed, which not only guaranteed the quality of the surface reconstruction, reduced the calculating time effectively, but also improved the reconstruction efficiency, strengthened the applicability of the 3D reconstruction method. The reconstructed 3D model of the femoral head cartilage included both the shape features and the thickness information of the cartilage, which could possess important reference value to medical diagnosis, rehabilitation monitoring and many treatment methods, such as cartilage replacement.