基于物质灰度正态分布的CT断层图像匹配插值

基于医学CT断层图像中物质灰度呈正态分布的特性,提出一种新的插值算法。先分割CT断层图像,再为每个插值点搜索属于同种物质的最佳匹配点对,最后基于物质灰度呈正态分布特性获取插值权值,计算出插值点的灰度值。针对灰度值相差较大的匹配点对,传统的线性插值方法容易生成误差,新算法避免了这些插值点误差的产生。对于存在较多灰度值差异大的匹配点对的4组测试切片,测试结果表明,与匹配点对间传统的线性插值方法相比较,新算法的标记点均偏差平均降低29.55%,说明新算法具有良好的插值效果。     

二维分形维数在皮肤癌光学相干断层成像技术诊断中的应用

目的:在利用光学相干断层成像技术检测和诊断皮肤癌时,获取能够用于皮肤癌诊断的图像特征值。方法:应用差分盒子技术方法,提取色素痣、基底细胞癌和黑色素瘤的光学相干断层图像的二维分形维数。结果:从光学相干断层图像中提取出的二维分形维数能够有效地区分色素痣、基底细胞癌及黑色素瘤,色素痣和基底细胞癌的分形维数较黑色素痣小,具有统计学意义。结论:应用本方法计算的二维分形维数可以用于皮肤癌的诊断。     

基于微机的医学图像三维重建

本文对由二维断层图像序列重建人体器官三维结构在微机上的实现进行了研究，先将目标物体从断层图像中分割出来，然后运用基于形状的目标插值形成二值化体数据，避免了线性灰度插值造成的物体表面边界模糊，并且缩短了预处理时间，采用一种改进的扫描线编码方法对体数据进行压缩，在微机有限的内存资源上实现了物体快速旋转和多种方式显示。     

基于计算统一设备架构的高斯径向基图像插值快速实现方法研究

在医学图像的处理及分析中,常常需要对图像进行插值运算。尽管高斯径向基(GRBF)插值有插值精度高的优点,但运算时间长的不足仍限制了它在图像插值中的应用。因此,本文提出采用基于计算统一设备架构(CUDA)的方法实现二维和三维医学图像的GRBF快速插值。根据CUDA单指令多线程(SIMT)的执行模型,采用合并访存、共享内存等各种合适的内存优化措施。并且在应用对数据空间进行二维分块,三维分体策略的过程中使用基于重叠区域的自然缝合算法来消除图像插值连接边界的失真现象。在保持较高图像插值精度的基础上,二维和三维医学图像GRBF插值各基本计算步骤都得到了极大的加速。实验结果表明:基于CUDA平台的GRBF插值执行效率与传统CPU运算相比明显提高,对其在图像插值中的应用具有相当的参考价值。     

基于改进支持向量机算法的超声图像分割技术

超声图像的边缘分割受到噪声影响,基于传统支持向量机(support vector machine,SVM)超声图像分割过程存在较大缺陷。提出一种基于改进SVM算法超声图像分割算法。采用分区域特征匹配方法,进行二维超声图像的分块融合性检测和特征块匹配,根据超声纹理的规则性特征分量进行病理边缘特征提取,利用提取的精度作为约束条件,优化SVM分割过程,进行超声图像分割过程的自适应分类,实现对超声图像的快速分割。仿真结果表明,采用该方法进行超声图像分割的精度较高,对超声图像的病理特征识别能力较好,结构相似度信息较强,提高了超声图像检测和诊断分析能力。     

医学三维成像内插技术的研究

由有限的二维断层图象系列获得三维图象，内插过程是必不可少的。本文对常用的两种内插－线性插值和动态弹性插值进行了实验的对比和分析。     

基于非刚体配准的多分辨率B样条层间插值方法

三维图像的处理和操作需要将一般的断层序列插值成为具有各坐标轴一致的分辨率的体数据。而目前最常用的线性插值方法在层间距较大时会导致图像边缘模糊和出现伪影。Penney根据现有的非刚体匹配方法,提出了利用图像形变场数据的插值算法,明显提高了层间插值的质量。本研究对Penney提出的算法进行了改进,用多分辨率B样条拟合插值方法替代Penney使用的最邻近直线插值方法,并将新算法的实验结果与原算法、线性插值进行了对比。新算法既保持了高质量插值又提高了计算速度,并且使生成的图像达到了C^2平滑。此算法可用于精度要求比较高的体数据插值重建过程。     

基于对应点匹配的断层图像三维插值方法

用于放疗计划的ＣＴ和ＭＲＩ图像断导蝗距离通常大于断层内部像素间的距离,三维剂量场的计算等工作通常需要等间隔分布的三维图像数据。目前常用的断层间直接线性值的方法分引起图像边界模糊,针对这个问题,本文提出了一种基于相邻两个断层图像对应点匹配的插值新方法,首先在相邻断层间建立起点与点的对应关系,然后再利用这些对应点插值出新的图像数据。模拟结果表明,客中方法能有效地消除直接线性插值带来的图像边界模糊的问题     

基于光学相干断层成像技术量化评估皮肤癌的二维傅里叶分形维数

光学相干断层成像技术(optical coherence tomography,OCT)可用于皮肤组织的诊断。从OCT图像提取的定量特征变量能够用作皮肤肿瘤诊断的指标。在这些定量特征变量中,分形维数能够表征皮肤组织的结构变化,从而能够提供有效的方法区分诊断皮肤肿瘤。本研究应用OCT获取了恶性黑色素瘤,基底细胞癌和色素痣OCT图像;基于傅里叶分析,从OCT图像中提取了二维分形维数;应用统计分析方法,分析各皮肤组织的分形维数的差异性。研究结果表明,与恶性黑素瘤相比,基底细胞癌和色素痣中的二维分形维数的显着减少;二维傅里叶分析能够提供一种简单易行的方法从OCT图像中提取分形维数,用以表征恶性黑色素瘤、基底细胞癌和色素痣。     

医学图像多维重建中的插值问题研究进展

插值广泛应用于医学成像和图像多维重建中。本文首先对传统插值方法中基于形状的插值和弹性匹配插值作了比较深入的介绍，然后分析了旋转扫描超声心脏图像插值方法的特点、难点和研究现状，论述了插值和匹配的关系，分析表明准确的旋转扫描插值方法应该是基于匹配的方法，最后讨论了图像插值的几种评价方法。     

Adaptive Voxel Matching for Temporal CT Subtraction

Temporal subtraction (TS) technique calculates a subtraction image between a pair of registered images acquired from the same patient at different times. Previous studies have shown that TS is effective for visualizing pathological changes over time; therefore, TS should be a useful tool for radiologists. However, artifacts caused by partial volume effects degrade the quality of thick-slice subtraction images, even with accurate image registration. Here, we propose a subtraction method for reducing artifacts in thick-slice images and discuss its implementation in high-speed processing. The proposed method is based on voxel matching, which reduces artifacts by considering gaps in discretized positions of two images in subtraction calculations. There are two different features between the proposed method and conventional voxel matching: (1) the size of a searching region to reduce artifacts is determined based on discretized position gaps between images and (2) the searching region is set on both images for symmetrical subtraction. The proposed method is implemented by adopting an accelerated subtraction calculation method that exploit the nature of liner interpolation for calculating the signal value at a point among discretized positions. We quantitatively evaluated the proposed method using synthetic data and qualitatively using clinical data interpreted by radiologists. The evaluation showed that the proposed method was superior to conventional methods. Moreover, the processing speed using the proposed method was almost unchanged from that of the conventional methods. The results indicate that the proposed method can improve the quality of subtraction images acquired from thick-slice images.     

基于改进光线投射法的冠脉图像三维重建

传统光线投射算法虽然绘制图像时能取得良好的效果，但因为运算速度较慢，限制了其广泛应用。为了在图像三维重建时既能保证质量，又能提高实时交互性，本文提出一种基于改进光线投射算法的冠脉三维重建方法，利用检测原数据点的数据值来简化重采样中三线性插值的复杂性，并结合切比雪夫空体素跳跃法，进一步提高算法的效率。实验结果表明，该算法能够在提高运算速度的同时，保证绘制冠脉三维图像的质量。     

Wobbled splatting--a fast perspective volume rendering method for simulation of x-ray images from CT

3D/2D registration, the automatic assignment of a global rigid-body transformation matching the coordinate systems of patient and preoperative volume scan using projection images, is an important topic in image-guided therapy and radiation oncology. A crucial part of most 3D/2D registration algorithms is the fast computation of digitally rendered radiographs (DRRs) to be compared iteratively to radiographs or portal images. Since registration is an iterative process, fast generation of DRRs-which are perspective summed voxel renderings-is desired. In this note, we present a simple and rapid method for generation of DRRs based on splat rendering. As opposed to conventional splatting, antialiasing of the resulting images is not achieved by means of computing a discrete point spread function (a so-called footprint), but by stochastic distortion of either the voxel positions in the volume scan or by the simulation of a focal spot of the x-ray tube with non-zero diameter. Our method generates slightly blurred DRRs suitable for registration purposes at framerates of approximately 10 Hz when rendering volume images with a size of 30 MB.     

Three-dimensional velocity mapping of lung motion using vessel bifurcation pattern matching

We present a new quantification technique for three-dimensional (3D) lung motion by means of tracking the anatomical features inside the lung using a set of sequential 3D-CT images (a 4D-CT image). The method is based on the conservation of topology, such as connections and junctions of vessels, during the motion. Lung CT images are used to do lung volume modeling, lung vessel extracting and thinning, and coordinates of vessel bifurcations are derived as feature points. Such feature points are tracked in a series of 3D-CT images, i.e., the points are individually tracked between two successive 3D-CT images, in which the lung is deformed. Consequently, 3D displacement vectors are obtained. The feature point tracking is carried out using point pattern matching with a probabilistic relaxation method. We examined this technique using a lung 3D-CT image and artificially deformed one, and separately scanned CT images for a rigid bifurcation phantom. The studies estimated that the error of the vectors is within approximately 1 voxel, i.e., 1 mm or less. Therefore, the accuracy is expected to be high enough for radiation therapy. This technique enables us to quantify realistic 3D organ motion without any fiducial markers. It can be applied to the quantification of tumor (target volume) deformation by gridding interpolation into all voxels. We expect it to be useful for dose estimation in mobile organs and for 4D treatment planning in radiation therapy.     

In vitro 3D reconstruction of long bones using B-scan image processing

An echographic image processing method has been developed, and validated by in vitro experiments, for the 3D reconstruction of the long bones of the newborn. The reconstruction of successive parallel cross-sections is obtained by a 2D reconstruction technique using radial B-scan image processing. The automatic segmentation of all the calculated images allows the extraction of the external contours of the skeleton. After structuring the explored volume using a contour association method, a contour interpolation step is required to solve the anisotropy problem, to obtain a 3D representation with cubic voxel lists. The results are encouraging, and a new mechanical part prototype of the acquisition system is under test for in vivo experiments. The main originality of the paper lies in the combination of different steps to obtain a practical solution to a clinical problem.     

Fast perspective volume ray casting method using GPU-based acceleration techniques for translucency rendering in 3D endoluminal CT colonography

Recent advances in graphics processing unit (GPU) have enabled direct volume rendering at interactive rates. However, although perspective volume rendering for opaque isosurface is rapidly performed using conventional GPU-based method, perspective volume rendering for non-opaque volume such as translucency rendering is still slow. In this paper, we propose an efficient GPU-based acceleration technique of fast perspective volume ray casting for translucency rendering in computed tomography (CT) colonography. The empty space searching step is separated from the shading and compositing steps, and they are divided into separate processing passes in the GPU. Using this multi-pass acceleration, empty space leaping is performed exactly at the voxel level rather than at the block level, so that the efficiency of empty space leaping is maximized for colon data set, which has many curved or narrow regions. In addition, the numbers of shading and compositing steps are fixed, and additional empty space leapings between colon walls are performed to increase computational efficiency further near the haustral folds. Experiments were performed to illustrate the efficiency of the proposed scheme compared with the conventional GPU-based method, which has been known to be the fastest algorithm. The experimental results showed that the rendering speed of our method was 7.72 fps for translucency rendering of 1024×1024 colonoscopy image, which was about 3.54 times faster than that of the conventional method. Since our method performed the fully optimized empty space leaping for any kind of colon inner shapes, the frame-rate variations of our method were about two times smaller than that of the conventional method to guarantee smooth navigation. The proposed method could be successfully applied to help diagnose colon cancer using translucency rendering in virtual colonoscopy.     

三维几何建模与仿真技术在牙颌修复中的应用

提出正常人牙颌组织从CT数据处理到目标组织几何建模及仿真设计的技术路线.将CT断层图像经过线性插值构造出三维数据场,分割出目标牙颌组织后,利用Marching Cubes算法提取牙颌组织的三角面几何模型.采用层切法重建完整的牙列咬和面,其层切精度达到0.2mm.经过配准,实现层切法和CT扫描所得图形线框的坐标拟合.然后在PowerSHAPE中对牙列单元进行三维重建,形成一套易于修改、方便组合的模型库;最后通过编程建立了牙列缺损与修复的几何仿真系统.     

基于运动估计的肺4D-CT图像冠矢状面超分辨率重建

肺4D-CT在肺癌放射治疗中发挥着重要的作用,但肺4D-CT数据层间的分辨率低,导致每个相位3D数据的肺冠矢状面均为低分辨率图像。本研究提出一种基于运动估计的超分辨率重建技术,以提高3D数据的冠矢状面图像分辨率。首先,分析图像退化模型;然后,采用基于完全搜索块匹配的运动估计法,估计出不同“帧”肺冠矢状面图像之间的运动场;最后,以此运动场为基础,采用迭代反投影法（IBP）,重建高分辨率的肺部冠矢状面图像。使用一个公共可用的数据集来评价所提出的算法,该数据集由10组肺4D-CT数据组成,每组数据包含10个相位。在每组图像中,选取不同相位的冠矢状面图像进行实验。结果表明,与传统的插值方法（如最近邻插值、双线性插值法）相比,图像边缘宽度均显著降低（最近邻插值9.93±0.59,双线性插值8.04±0.69,新算法5.41±0.60, P<0.001）;较双线性插值,图像平均梯度显著提高（5.41±0.59 vs 7.49±0.75, P<0.001）,新方法不仅能获得视觉上清晰的图像,而且量化评价指标也有明显提高。主观和客观实验结果表明,所提出的新方法能有效提高肺4D-CT冠矢状面图像的分辨率。     

Dynamic ventilation imaging from four-dimensional computed tomography

A novel method for dynamic ventilation imaging of the full respiratory cycle from four-dimensional computed tomography (4D CT) acquired without added contrast is presented. Three cases with 4D CT images obtained with respiratory gated acquisition for radiotherapy treatment planning were selected. Each of the 4D CT data sets was acquired during resting tidal breathing. A deformable image registration algorithm mapped each (voxel) corresponding tissue element across the 4D CT data set. From local average CT values, the change in fraction of air per voxel (i.e. local ventilation) was calculated. A 4D ventilation image set was calculated using pairs formed with the maximum expiration image volume, first the exhalation then the inhalation phases representing a complete breath cycle. A preliminary validation using manually determined lung volumes was performed. The calculated total ventilation was compared to the change in contoured lung volumes between the CT pairs (measured volume). A linear regression resulted in a slope of 1.01 and a correlation coefficient of 0.984 for the ventilation images. The spatial distribution of ventilation was found to be case specific and a 30% difference in mass-specific ventilation between the lower and upper lung halves was found. These images may be useful in radiotherapy planning.     

Applying an animal model to quantify the uncertainties of an image-based 4D-CT algorithm

The purpose of this paper is to use an animal model to quantify the spatial displacement uncertainties and test the fundamental assumptions of an image-based 4D-CT algorithm in vivo. Six female Landrace cross pigs were ventilated and imaged using a 64-slice CT scanner (GE Healthcare) operating in axial cine mode. The breathing amplitude pattern of the pigs was varied by periodically crimping the ventilator gas return tube during the image acquisition. The image data were used to determine the displacement uncertainties that result from matching CT images at the same respiratory phase using normalized cross correlation (NCC) as the matching criteria. Additionally, the ability to match the respiratory phase of a 4.0 cm subvolume of the thorax to a reference subvolume using only a single overlapping 2D slice from the two subvolumes was tested by varying the location of the overlapping matching image within the subvolume and examining the effect this had on the displacement relative to the reference volume. The displacement uncertainty resulting from matching two respiratory images using NCC ranged from 0.54 ± 0.10 mm per match to 0.32 ± 0.16 mm per match in the lung of the animal. The uncertainty was found to propagate in quadrature, increasing with number of NCC matches performed. In comparison, the minimum displacement achievable if two respiratory images were matched perfectly in phase ranged from 0.77 ± 0.06 to 0.93 ± 0.06 mm in the lung. The assumption that subvolumes from separate cine scan could be matched by matching a single overlapping 2D image between to subvolumes was validated. An in vivo animal model was developed to test an image-based 4D-CT algorithm. The uncertainties associated with using NCC to match the respiratory phase of two images were quantified and the assumption that a 4.0 cm 3D subvolume can by matched in respiratory phase by matching a single 2D image from the 3D subvolume was validated. The work in this paper shows the image-based 4D-CT algorithm to be a promising method for producing 4D-CT images for radiotherapy.