A feature-based solution for 3D registration of CT and MRI images of human knee

Abstract

This paper presents a feature-based solution for 3D registration of CT and MRI images of a human knee. It facilitates constructing high-quality models with clear outlining of bone tissues and detailed illustration of soft tissues. The model will be used for analysing the effect of posterior cruciate ligament and anterior cruciate ligament deficiency. The solution consists of preprocessing, feature extraction, transformation parameter estimation and resampling, and blending. In preprocessing, we propose partial preserving and iterative neighbour comparing filtering to help segment bone tissues from MRI images without having to construct a statistical model. Through analysing the characteristics of knee images, tibia and femur are selected as the features and the algorithm for effectively extracting them is described. To estimate transformation parameters, we propose a method based on the statistical information of projected feature images, including translating according to the project feature image centroids and calculating the rotation angle by searching and mapping boundary points. We transform the MRI image and then blend it with the CT image by taking the maximum

Complex Dual Tree Wavelet Multi-Level Feature Based Transformation Parameter Estimation for 3d Medical Image Registration

“…In Image registration, the deformed images is transformed by a transformation factor of T that is computed by considering the distance between features such as edges, points and regions [4] that are present in the deformed and reference image. In [5] Registration of MRI and CT images by selection of tibia and femur features from the binary images and affine transformation, validation of the algorithm is carried out by considering correlation coefficient. In [6], preprocessing techniques such as geometric feature based segmentation dynamic threshold method, feature extraction techniques such as automated trunk slices extraction and transformation technique like multithread iterative closest point are used for registration.…”

“…Validation metrics such as negative normalization correlation and Euclidean distance error are computed for registering PET and CT images. In [5], phase information coefficients are considered to align images by using Dual Tree Complex wavelet transforms (DTCWT) coefficients. The shift invariance property of the front end filter along with directional selective property available in DTCWT bands the registration algorithm becomes robust to local mean and contrast changes in images.…”

“…Coarse level registration is carried out by considering edges and cross correlation function from low frequency bands and the registration is improved by considering mutual information from six orientation bands. In [8], registration of medical images is carried out by computing DTCWT sub bands for key point's selection, Hausdorff distance for similarity measurement between key points and Niche particle swarm optimization for affine transformation. Evaluation metrics such as correlation coefficient, normal mutual information, mean square error and time are used.…”

Fast Registration Algorithm for 3D Medical Images using Double Density Dual Tree Complex Wavelet Transform and Bi-directional Prediction

“…RANSAC is applied for outlier elimination and to robustly estimate the best fitting homography, and at end homography transformation is used as transformation model. In [6] Registration of MRI and CT images by selection of tibia and femur features from the binary images and affine transformation, validation of the algorithm is carried out by considering correlation coefficient. In [7], preprocessing techniques such as geometric feature based segmentation dynamic threshold method, feature extraction techniques such as automated trunk slices extraction and transformation technique like multithread iterative closest point are used for registration.…”

A Registration Method Based on Contour Point Cloud for 3D Whole-Body PET and CT Images

“…The study in [ 9 ] combined the segmentation and intensity information together for PET/CT registration, whose speed and accuracy were enhanced. As for feature-based methods, [ 10 ] presented a preprocessing method for MRI and CT images and a feature-based affine transformation where tibia and femur were selected as the feature for the registration of human knee images. Reference [ 11 ] modified Iterative Closest Points (ICP) by combining moments, center points, and Canny calculators, which had a good result on head images with a high speed.…”