Research Of Medical Image Registration Algorithms

Using images to guide therapeutic, such as surgical, radio-surgical, and radiotherapeutic, planning is a rapidly growing field. Precise registration would provide useful clinical information. It can increase treatment efficiency and minimize neurological damage. Given two image sets acquired from the same patient but at different times or with different devices, image registration is the process of finding the geometric transformation that aligns one image to another. The geometric alignment or registration of multimodality images is an essential and fundamental task to clinical diagnosis. X-ray computed tomography (CT) data depict high contrast structure of bony tissue, magnetic resonance imaging (MRI) and , Ultrasound Image (US) provide excellent contrast of soft tissue, and nuclear imaging techniques such as positron emission tomography (PET) and single photon emission computed tomography (SPECT) provide functional data. Analysis of the data is enhanced by the complementary information present in a multimodality study, and techniques exist that use anatomical datasets to further process functional images. In order to analyze or utilize data from two different modalities, accurate registration is necessary. Image registration is mainly applied to the areas of video compression and coding, motion analysis, objection tracking etc.We mainly focus our research on four image registration topics, as follows:1 Medical Image Registration Using Concurrence Mutual InformationMutual information (MI) is calculated on a pixel by pixel basis. It takes into account only the relationship between corresponding individual pixels rather than those of each pixel's respective neighborhood, which ignores the spatial information. In this article we propose a new measure—Concurrence Mutual Information(CMI). It is an extension of the mutual information framework which incorporates spatial information about image structure into the registration process, and has the potential to improve the accuracy and robustness of image registration. The results indicate that CMI is a more robust similarity measure for image registration than MI. Furthermore CMI can be used in other fields, such as economics, operational research and pattern recognition, just like MI.2 Medical Image Registration Based on Principal Phase CongruencyA novel image attribute, principal phase congruency (PPC), is defined and used to register medical images. Phase congruency is computed on different scales and orientations. PPC can be developed from a fusion of the phase congruency by using principal component analysis. A fuzzy similarity measure is introduced as the registration function. We evaluate the effectiveness of the proposed approach by applying it to the simulated and real brain image data (CT, MR and PET). Experimental results indicate that the algorithm is less sensitive to low sampling resolution and noise, does not contain incorrect local maxima that are sometimes found in the traditional algorithm.3 Fast 3-D Medical Image Registration Based on Equivalent Meridian PlaneAs is well known, a meridian plane is arbitrary plane perpendicular to the celestial equator, which passes through the earth's axis of rotation. For three-dimensional medical image, it is necessary to propose a new Equivalent Meridian Plane (EMP) concept since estimating the meridian plane is not always feasible in practice.For a three-dimensional irregular volume, a set of orthogonal principal axes can be always found, by which a family of orthogonal planes can be determined. One of these planes, containing the first and the second principal axis, is defined as EMP.For the rigid registration of multi-modality medical images, mutual information (MI) technique is unsuitable to clinical diagnose because of high computational cost and low robustness. A new concept of equivalent meridian plane (EMP) is proposed, and the EMP and other two normal feature planes are determined using principal component analysis (PCA); the rough registrations of those 2D planes are carried out at six freedom degree; finally, the refine registrations can be completed using MI in a small neighboring region. This method is called as EMP based MI registration technique. The accuracy and robustness of EMP-MI approach can be verified by applying it to the simulated and real brain image data (CT, MR, PET, and SPECT). The experimental results indicate that the proposed algorithm reduces computational time distinctly and is a global optimal strategy.4 Hybrids Rigid and Non-rigid Registration Algorithm for Alignment of Serial Thoracic and Abdominal ImagesRegistration of the same subject with the same modality at different time is useful for physicians, either to follow the development of a disease, or for interventions (dynamic acquisition during the operation or its validation).A hybrid rigid and non-rigid registration algorithm has been presented to register thoracic and abdominal CT images of the same subject scanned at different times. The bony structures are first segmented from two different time CT images, respectively. Then, the segmented bony structures in the two respective images are registered based on their boundary points using a soft correspondence matching algorithm with a rigid transformation constraint on each bony structure. With estimated correspondences in bony structures, the dense deformations in the entire images are interpolated by a Thin Plate Spline (TPS) interpolation technique. To improve the alignment of soft tissues in the images as well, a normalized mutual information based B-Spline registration algorithm is used to iteratively refine the registration of soft tissues, and at the same time keeps the rigid transformation for each bony structure. This registration refinement procedure is repeated until the algorithm converges. The proposed hybrid registration algorithm has been applied to the clinical data with encouraging results as evaluated by two clinical radiologists.