Medical Image Registration Based on Phase Congruency and Regional Mutual Information

In this paper, a new approach of muhi-modality image registration is represented with not only image intensity, but also features describing image structure. There are two novelties in the proposed method. Firstly, instead of standard mutual information （ MI ） based on joint intensity histogram, regional mutual information （ RMI ） is employed, which allows neighborhood information to be taken into account. Secondly, a new feature images obtained by means of phase congruency are invariants to brightness or contrast changes. By incorporating these features and intensity into RMI, we can combine the aspects of both structural and neighborhood information together, which offers a more robust and a high level of registration accuracy.

A method based on mutual information and gradient information for medical image registration

Mutual information is widely used in medical image registration, because it does not require preprocessing the image. However, the local maximum problem in the registration is insurmountable. We combine mutual information and gradient information to solve this problem and apply it to the non-rigid deformation image registration. To improve the accuracy, we provide some implemental issues, for example, the Powell searching algorithm, gray interpolation and consideration of outlier points. The experimental results show the accuracy of the method and the feasibility in non-rigid medical image registration.

Image Registration Based on Improved Mutual Information with Hybrid Optimizer

An improved image registration method is proposed based on mutual infor- mation with hybrid optimizer. Firstly, mutual information measure is combined with morphological gradient information. The essence of the gradient information is that locations a large gradient magnitude should be aligned, but also the orientation of the gradients at those locations should be similar. Secondly, a hybrid optimizer combined PSO with Powell algorithm is proposed to restrain local maxima of mutual information function and improve the registration accuracy to sub-pixel level. Lastly, muhlresolution data structure based on Mallat decomposition can not only improve the behavior of registration function, but also improve the speed of the algorithm. Experimental results demonstrate that the new method can yield good registration result, superior to traditional optimizer with respect to smoothness and attraction basin as well as convergence speed.

Robust Image Registration Based on Mutual Information Measure

A new implementation of the image registration algorithm based on the mutual information is presented for the case of medical images. The registration is achieved if the maximum of the mutual information is attained. In this maximization process optimal values of five parameters of an affine transformation are searched.

Study of three-dimensional PET and MR image registration based on higher-order mutual information

Mutual information has currently been one of the most intensivelyresearched measures. It has been proven to be accurate and effective registrationmeasure. Despite the general promising results, mutual information sometimes mightlead to misregistration because of neglecting spatial information and treating intensityvariations with undue sensitivity. In this paper, an extension of mutual informationframework was proposed in which higher-order spatial information regarding imagestructures was incorporated into the registration processing of PET and MR. Thesecond-order estimate of mutual information algorithm was applied to the registrationof seven patients. Evaluation from Vanderbilt University and our visual inspectionshowed that sub-voxel accuracy and robust results were achieved in all cases withsecond-order mutual information as the similarity measure and with Powell's multi-dimensional direction set method as optimization strategy.

Registration Method for CT-MR Image Based on Mutual Information

Medical image registration is important in many medical applications. Registration method based on maximization of mutual information of voxel intensities is one of the most popular methods for 3-D multi-modality medical image registration. Generally, the optimization process is easily trapped in local maximum, resulting in wrong registration results. In order to find the correct optimum, a new multi-resolution approach for brain image registration based on normalized mutual information is proposed. In this method, to eliminate the effect of local optima, multi-scale wavelet transformation is adopted to extract the image edge features. Then the feature images are registered, and the result at this level is taken as the initial estimate for the registration of the original images. Three-dimensional volumes are used to test the algorithm. Experimental results show that the registration strategy proposed is a robust and efficient method which can reach sub-voxel accuracy and improve the optimization speed.

Graphic Processing Unit-Accelerated Mutual Information-Based 3D Image Rigid Registration

Mutual information （MI）-based image registration is effective in registering medical images, but it is computationally expensive. This paper accelerates MI-based image registration by dividing computation of mutual information into spatial transformation and histogram-based calculation, and performing 3D spatial transformation and trilinear interpolation on graphic processing unit （GPU）. The 3D floating image is downloaded to GPU as flat 3D texture, and then fetched and interpolated for each new voxel location in fragment shader. The transformed resuits are rendered to textures by using frame buffer object （FBO） extension, and then read to the main memory used for the remaining computation on CPU. Experimental results show that GPU-accelerated method can achieve speedup about an order of magnitude with better registration result compared with the software implementation on a single-core CPU.

Fast Mutual Information Registration Method of 3-D Medical Image

Currently the voxel based registration methods have been used widely such as the well known mutual information (MI). Although the accuracy of their results is plausible, the registration procedure is slow. This paper proposed some methods to rigid registration based on mutual information, aiming for an acceleration of the registration process without significantly loss of precision in the final results. The efficiency of these methods is examined by registration of CT MR and PET MR. Experimental results show that the speedup is effective and efficient. By using the fast methods, the registration of 3 D medical image could also be implemented on PC rapidly.

A new hybrid particle swarm optimization for multimodal brain image registration

Image registration is an important issue in medical analysis. In this process the spatial transformation that aligns the reference image and the floating image is estimated by optimizing a similarity metric. Mutual information (MI), a popular similarity metric, is a reliable criterion for medical image registration. In this paper, we present an improved method for multimodal image registration based on maximization of a new form of normalized MI incorporating particle swarm optimization, PSO, as a searching strategy. Also a new hybrid PSO algorithm is applied to approach more precise and robust results with better performance.

MUTUAL INFORMATION BASED 3D NON-RIGID REGISTRATION OF CT/MR ABDOMEN IMAGES

A mutual information based 3D non-rigid registration approach was proposed for the registration of deformable CT/MR body abdomen images. The Parzen Windows Density Estimation (PWDE) method is adopted to calculate the mutual information between the two modals of CT and MRI abdomen images. By maximizing MI between the CT and MR volume images, the overlapping part of them reaches the biggest, which means that the two body images of CT and MR matches best to each other. Visible Human Project (VHP) Male abdomen CT and MRI Data are used as experimental data sets. The experimental results indicate that this approach of non-rigid 3D registration of CT/MR body abdominal images can be achieved effectively and automatically, without any prior processing procedures such as segmentation and feature extraction, but has a main drawback of very long computation time.

Mutual Information Optimization Based Dynamic Log-Polar Image Registration

Log-polar transformation(LPT)is widely used in image registration due to its scale and rotation invariant properties.Through LPT,rotation and scale transformation can be made into translation displacement in log-polar coordinates,and phase correlation technique can be used to get the displacement.In LPT based image registration,constant samples in digitalization processing produce less precise and effective results.Thus,dynamic log-polar transformation(DLPT)is used in this paper.DLPT is a method that generates several sample sets in axes to produce several results and only the effective results are used to get the final results by using statistical approach.Therefore,DLPT can get more precise and effective transformation results than the conventional LPT.Mutual information(MI)is a similarity measure to align two images and has been used in image registration for a long time.An optimal transform for image registration can be obtained by maximizing MI between the two images.Image registration based on MI is robust in noisy,occlusion and illumination changing circumstance.In this paper,we study image registration using MI and DLPT.Experiments with digitalizing images and with real image datasets are performed,and the experimental results show that the combination of MI with DLPT is an effective and precise method for image registration.

A MNCIE method for registration of ultrasound images

A new approach to the problem of registration of ultrasound images is presented, using a concept of Nonlinear Correlation Information Entropy (NCIE) as the matching criterion. The proposed method applies NCIE to measure the correlation degree between the image intensities of corresponding voxel in the floating and reference images. Registration is achieved by adjustment of the relative position until NCIE between the images is maximized. However, unlike mutual information (MI), NCIE varies in the closed interval [0, 1], and around the extremum it varies sharply, which makes it possible that thresholds of NCIE can be used to boost the search for the registration transformation. Using this feature of NCIE, we combine the downhill simplex searching algorithm to register the ultrasound images. The simulations are conducted to testify the effectiveness and rapidity of the proposed registration method, in which the ultrasound floating images are aligned to the reference images with required registration accuracy. Moreover, the NCIE based method can overcome local minima problem by setting thresholds and can take care of the differences in contrast between the floating and reference images.

COMPACT SUPPORT THIN PLATE SPLINE ALGORITHM

Common tools based on landmarks in medical image elastic registration are Thin Plate Spline (TPS) and Compact Support Radial Basis Function (CSRBF). TPS forces the corresponding landmarks to exactly match each other and minimizes the bending energy of the whole image. However, in real application, such scheme would deform the image globally when deformation is only local. CSRBF needs manually determine the support size, although its deformation is limited local. Therefore, to limit the effect of the deformation, new Compact Support Thin Plate Spline algorithm (CSTPS) is approached, analyzed and applied. Such new approach gains optimal mutual information, which shows its registration result satisfactory. The experiments also show it can apply in both local and global elastic registration.

基于互信息量均方差提取关键帧的激光视频图像检索研究

为保证激光视频图像检索结果中不存在重复性冗余图像,提出了基于互信息量均方差提取关键帧的激光视频图像检索方法。基于互信息量均方差的关键帧提取方法,以激光视频图像颜色的互信息量均方差最大化,为激光视频图像关键帧的聚类中心设置标准,以此聚类提取不重复的视频图像关键帧;通过基于关键帧的激光视频图像检索方法,将所提取关键帧作为激光视频图像检索的核心判断内容,提取与所需图像关键帧相似度显著的激光视频图像,完成激光视频图像检索。实验结果显示:此方法使用后,提取的激光视频图像关键帧冗余度仅有0.01,激光视频图像检索结果的MAP指标测试值高达0.98,检索结果中不存在重复性冗余图像。

基于互信息与萤火虫算法的网络入侵特征选择

为减少网络入侵检测数据中的冗余特征,提出一种结合互信息和萤火虫算法的特征选择方法。针对互信息不能精确计算特征间冗余度,提出类内特征冗余互信息特征选择方法。针对萤火虫算法步长因子固定易使算法陷入局部最优等问题,提出自适应步长萤火虫算法特征选择。以上方法分别选取特征子集后利用投票策略选取最优子集,对该子集基于C4.5和贝叶斯网络分类器分类。实验结果表明,使用10个特征检测能有效提高入侵检测率、误报率和F-measure,同时还缩短训练和检测时间。此外,与现有的几种方法相比,该方法在准确率、检测率和F-measure都获得不错效果。

融合改进头脑风暴与Powell算法的马铃薯多模态图像配准

基于热成像仪获取作物冠层温度可以实现作物水分胁迫状态的非接触式、无损检测,并且具有高通量检测的潜力。然而热红外图像存在作物边缘分布不清晰、噪声强、缺乏形状、纹理信息等问题,无法实现作物冠层温度自动化提取,利用可见光与热红外图像间的信息互补性,通过图像自动配准技术可以弥补热红外图像缺点,为自动化检测提供基础。为解决可见光图像与热红外图像之间辐射、形状和纹理差异,导致不同模态图像配准难度较大问题,该研究提出了一种融合改进头脑风暴(brain storm optimization algorithm,BSO)与Powell算法的可见光与热红外图像配准方法。研究通过对原始BSO优化算法进行改进使得整体算法更好寻找到最优仿射变换矩阵进而完成图像配准任务,具体改进包含以下5个方面:使用混沌映射函数初始化BSO群体分布、修改新个体变异范围、手肘法动态调整BSO中K-means聚类数、在个体变异方式策略中加入混沌本地搜索方法、在算法执行过程中根据BSO算法前期后期不同特性动态调整概率参数。研究选用互信息值(mutual information,MI)、归一化互信息值(normalized mutual information,NMI)、均方根误差(root mean square error,RMSE)和平均结构相似性指数(mean structure similarity index measure,MSSIM)作为评价指标。该研究算法相对比Powell优化算法、遗传算法(genetic algorithm,GA)和BSO_Powell算法在温室数据中MI指标分别提升0.054 2、0.076 9、0.040 5,NMI指标分别提升0.015 9、0.023 1、0.052 7,RMSE指标分别降低15.02、13.03、27.08,MSSIM指标分别提升0.052 3、0.048 8、0.122 4;大田数据中MI指标分别提升0.064 2、0.066 7、0.035 5,NMI指标分别提升0.007 7、0.012 5、0.012 4,RMSE指标分别降低14.06、10.57、15.40,MSSIM指标分别提升0.047 1、0.038 1、0.042 9。结果表明,所提出算法具有很强的鲁棒性,能够准确完成复杂环境下马铃薯多模态图像配准任务。

X线图像引导系统KylinRay-XGPS关键技术研发

图像引导用以实现患者摆位的精确定位,是放射治疗等领域实现精准治疗的重要保障。常见的锥形束CT(CBCT)图像引导设备花费时间长且辐射剂量大。基于此,本研究基于X线透视成像技术开发了图像引导系统KylinRay-XGPS。该系统根据采集的X线图像和患者定位CT进行2D-3D图像配准,输出患者配准平移和旋转误差,从而纠正患者的摆位误差。KylinRay-XGPS系统的关键技术有动态步长法生成数字影像重建图像(DRR)、基于GPU并行加速、多模态互信息灰度配准和基于骨性结构的配准方法。结合KylinRay-XGPS在放射治疗的临床病例测试结果表明,KylinRay-XGPS具有定位精度高、定位时间短和成像剂量低的优势。

基于纹理增强和细化的可见光图像互信息边缘提取研究

采用传统方法提取图像边缘时,会产生图像互信息匹配不准确、F_(1)值低、边缘检测性能不理想等问题。基于此,提出基于纹理增强和细化的可见光图像互信息边缘提取的方法:利用互信息配准公式对图像进行配准,并在此基础上对图像进行预处理,处理步骤包括图像纹理增强、图像二值化、图像平滑及细化,最终得到无干扰的清晰数据,再利用二维光滑化函数和Lipschitz方法对优化后的图像进行边缘检测和提取。试验结果表明,所提方法的F_(1)值始终>0.85,在3组试验数据集中的真正例率>95%,且假正比例<20%。试验所得数据说明所提方法具有更好的应用性能。

基于萤火虫算法的互信息医学图像配准

针对互信息配准方法中目标函数因存在多极值而容易陷入局部最优的问题,提出了一种基于萤火虫算法改进优化策略的互信息医学图像配准算法。该算法使用归一化互信息作为相似性测度,用萤火虫所处位置来表示配准参数,根据每个萤火虫的位置计算互信息函数值并将其作为当前萤火虫的亮度,通过亮度和吸引度的迭代更新来寻找互信息函数取最优解时的最佳配准参数。实验结果表明,该方法克服了互信息函数容易陷入局部最优的问题,有效地提高了配准精度。

基于互信息的医学图像配准技术研究进展

基于互信息的图像配准方法具有自动化程度高、配准精度高等优点,已被广泛应用于医学图像的配准。详细地介绍了互信息的基本概念、归一化互信息、互信息计算中常用的变换模型、插值方法、优化算法及提高配准速度策略等。同时对互信息在配准技术中存在的问题及解决方法进行了系统地介绍,最后对今后互信息的研究工作作了展望。