医学图像体绘制中的快速三线性插值算法

目的 三线性插值是医学图像体绘制中的基本运算单元，每次采样后都需要进行，因此高效快速的采样计算是提高体绘制成像速度的重要途径之一，特别是在微型机上实现医学图像的体绘制。本文目的是提出了一种全新的快速三线性插值算法。方法 根据体数据中单个体素的8个顶点的数值分布，把全部体素分成128类，并用一个字节M中的各位来表示，插值计算时可根据每种体素的M值来选择相应的插值计算公式，从而极大地减少了插值计算的总量。此外，通过分类时的阈值设定，还可以灵活地改变三线性插值的运算总量。结果 针对体数据的特性，提出了一种高效灵活的三线性插值算法。结论 与其他快速三线性插值算法比较，该算法不仅能够显著减少计算量，还可以根据对结果图像的精度要求，通过设定的阈值对体数据的单个体素进行分类，灵活地调整三线性插值的运算总量，提高体绘制的速度。     

采用半透明效果实现医学三维整体可视化

目的通过改进直接体绘制技术（DVR）算法中的关键步骤，绘制能表达数据场中不同层次组织的内部结构。方法根据界面对光线的反射和物质本身对光线的衰减，将数据场划分为边界不丰富的简单数据场和边界丰富的复杂数据场，从而采用分段线性函数来构造阻光度传递函数（TF）；将数据场中低灰度体素点看为一种不发光但会导致采样光线衰减的物质；根据采样结束时采样光线的阻光度分布来调整TF，完成三维绘制。结果绘制出了不同层次组织的内部结构。结论本算法有效地解决了体素点相互遮挡的问题，实现了医学三维整体可视化。     

医学体视化的三维旋转绘制法

目的 实现三维医学图像数据场的快速旋转体绘制。方法 由源一衰减模型、朗伯漫反射余弦定律和Phong镜面反射模型，推导出计算数据场中每一点的不透明度和明暗度的计算公式,通过旋转光源和视点,来自不同观察方向的感兴趣结构的体绘图被满意地生成。结果 通过对人体头部CT图像数据场进行三维旋转绘制，结果比较理想。结论 由于不需要对整个数据场进行分类和旋转，该方法具有速度快、所需内存少和绘制质量高的特点，是一种较好的医学体视化方法。     

激光共聚焦显微镜数据的交互体绘制算法

目的在PC机上实现激光共聚焦显微镜数据的实时交互体绘制 ,以满足用户对数据场的不同观察要求。方法通过分析激光共聚焦显微镜数据光学属性 ,得到交互传递函数 ,使用户可以改变绘制参数来参与对激光共聚焦显微镜数据场的数据挖掘 ;同时采用了纹理映射体绘制技术进行实时三维重建。结果在普通PC上实现了本算法。对一空间分辨率为 2 5 6× 2 5 6× 40的激光共聚焦显微镜数据场进行了三维重建实验 ,得到一组满足用户不同观察目的的重建结果。重建过程中 ,在用户更改绘制参数时 ,可以 1 0帧 /s的速度进行重绘制。结论与传统算法不同 ,本算法可在PC上提供给用户实时交互环境 ,更方便地获得符合观察要求的重建结果。     

基于表面点的高分辨率体数据三维重建

目的对高分辨率体数据实现高质量的实时三维表面浏览。方法本算法用结合阈值和形态学的分割方法提取体数据中单个器官的三维表面点集,再根据体数据中灰度梯度得到表面点的法向量。通过用表面点代替三角形面片来描述器官表面,在用户定义器官表面的颜色和透明度后,利用显卡OpenGL接口对表面点集进行三维显示。结果在微机环境下对中国数字人男性一号的CT数据集的骨骼和体表进行了三维重建,在保证图像质量的前提下重建速度超过25帧/s。结论本文提出的三维表面重建算法能对512×512×1720的高分辨率体数据进行高质量的实时三维表面浏览。     

基于PC机的虚拟内窥镜成像算法

目的对现有的虚拟内窥镜成像方法进行改进 ,使之能够在PC机上运行。方法传统算法在观察点位置变化时 ,应先对所有原始体数据进行三维坐标变换后再作透视投影 ,本文算法只对被观察的那部分数据进行处理 ,减少不必要的坐标变换 ,并将坐标变换、表面检测、透视投影等紧密结合 ,在一次光线跟踪过程中完成这些计算以减小计算量 ;直接对原始体数据进行投影 ,无需先对体数据进行插值 ,因此不用保留插值的中间结果 ,减少了内存消耗 ;在成像过程中自动排除与成像结果无关的数据 ,使它们不参与插值 ,不参与投影 ,进一步提高了计算速度。结果设计出基于PC机的虚拟内窥镜系统 ,并用大小为 5 1 2× 5 1 2的CT图片 ,绘制了胸部气管的内窥图像 ,以验证本文算法的有效性。结论所提出的算法能够在PC机上实现 ,有助于虚拟内窥镜系统的推广应用     

一种高效的光线投射体绘制优化算法

目的针对传统光线投射体绘制算法绘制速度慢,绘制图像有木纹效应的问题,提出一种高效的光线投射体绘制优化算法,消除绘制图像的木纹效应,同时提高体绘制的速度。方法采用预积分分类方法避免采样频率对传递函数的依赖,消除光线投射体绘制产生的木纹效应,利用平行线切割投射光线的数学特性实现高效的插值方法,提高体绘制的速度。利用图像的均方误差MSE和峰值信噪比PSNR评价图像的质量。结果本文的优化算法不但能够消除图像的木纹效应,而且使体绘制速度提高2.4倍以上。结论本文的光线投射体绘制优化算法能抑制图像的木纹效应,得到高质量图像,并且提高体绘制的速度。     

基于体素相似性的三维多模脑图像配准研究

为了对不同模态下的三维脑图像进行配准研究,引入一种基于体素相似性的配准方法。深入研究三维空间变换模型与三维配准体数据,并分析了互信息算法陷入局部极值的原因;本研究使用具有多分辨率策略的Mattes互信息算法对三维多模脑图像进行配准,通过对多模三维体图像进行三维单模和多模配准发现这种混合算法精度高,鲁棒性强,有效地降低了测度函数在收敛过程中陷入局部最优的可能性,所有的配准误差都小于一个像素大小,配准的精度达到亚像素级标准。即加入多分辨率策略的Mattes互信息算法较好地提高了三维多模脑图像配准的速度、精度和鲁棒性。     

体素形态学方法与人工神经网络结合在Alzheimer病诊断中的应用研究

目的 将体素形态学方法与人工神经网络方法相结合,用于阿尔茨海默氏病的诊断,探索自动、客观的Alzheimer临床诊断新途径.方法 研究对象包括10名可能的Alzheimer病人和12例年龄与性别与之相匹配的正常老年人,在Siemens Sonata 1.5T超导MR成像系统上获得精细的三维结构图像,首先利用体素形态学方法得到了Alzheimer患者灰质发生缺失的10个区域,然后统计每个区域的灰度值,在此基础上,利用反向传播神经网络对数据进行分析处理.结果 海马、海马旁回、内嗅皮层、杏仁体、尾状核头部、颞中回、扣带回、顶下小叶、岛叶和前额背外侧区域,在训练样本数为18的情况下,利用神经网络能够达到100%的判别效果.结论 将人工神经网络方法和体素形态学(VBM)方法相结合诊断Alzhei-mer患者,具有良好的临床应用前景,可能成为一种实用和可靠的临床诊断工具.     

基于体素的MRI形态分析诊断Alzheimer病的价值

目的评价基于体素的MRI形态分析（voxel-based Morphometry，VBM）法诊断Alzheimer病（AD）的价值。方法采用Siemens Sonata1．5T超导MR成像系统，对19例可能AD患者和15例年龄、性别与之相匹配的健康老年人及15例正常年轻人行全脑扫描，应用磁化准备快速梯度回波成像序列获取三维结构图像。数据分析采用以SPM99软件包为基础的全自动VBM技术进行处理。结果与健康老年对照组比较，AD患者两侧内颞叶（海马）明显萎缩（P〈0．001），右侧海马减少的体素总数为1529个，左侧海马减少的体素总数为1281个；而且AD患者右侧尾状核头和左侧内侧丘脑亦明显萎缩，减少的体素总数均为1529个；而感觉运动皮层、枕叶及小脑相对保持完好。此外，AD患者大脑皮层萎缩具有不对称性。结论VBM法操作简单，可自动化观察AD全脑萎缩情况，所得结论不仅与既往神经影像学研究结果一致，并得到了有重要价值的新发现，具有广阔的临床应用前景。     

Automatic transfer function design for medical visualization using visibility distributions and projective color mapping

Transfer functions play a key role in volume rendering of medical data, but transfer function manipulation is unintuitive and can be time-consuming; achieving an optimal visualization of patient anatomy or pathology is difficult. To overcome this problem, we present a system for automatic transfer function design based on visibility distribution and projective color mapping. Instead of assigning opacity directly based on voxel intensity and gradient magnitude, the opacity transfer function is automatically derived by matching the observed visibility distribution to a target visibility distribution. An automatic color assignment scheme based on projective mapping is proposed to assign colors that allow for the visual discrimination of different structures, while also reflecting the degree of similarity between them. When our method was tested on several medical volumetric datasets, the key structures within the volume were clearly visualized with minimal user intervention.     

Dynamic real-time 4D cardiac MDCT image display using GPU-accelerated volume rendering

Intraoperative cardiac monitoring, accurate preoperative diagnosis, and surgical planning are important components of minimally-invasive cardiac therapy. Retrospective, electrocardiographically (ECG) gated, multidetector computed tomographical (MDCT), four-dimensional (3D + time), real-time, cardiac image visualization is an important tool for the surgeon in such procedure, particularly if the dynamic volumetric image can be registered to, and fused with the actual patient anatomy. The addition of stereoscopic imaging provides a more intuitive environment by adding binocular vision and depth cues to structures within the beating heart. In this paper, we describe the design and implementation of a comprehensive stereoscopic 4D cardiac image visualization and manipulation platform, based on the opacity density radiation model, which exploits the power of modern graphics processing units (GPUs) in the rendering pipeline. In addition, we present a new algorithm to synchronize the phases of the dynamic heart to clinical ECG signals, and to calculate and compensate for latencies in the visualization pipeline. A dynamic multiresolution display is implemented to enable the interactive selection and emphasis of volume of interest (VOI) within the entire contextual cardiac volume and to enhance performance, and a novel color and opacity adjustment algorithm is designed to increase the uniformity of the rendered multiresolution image of heart. Our system provides a visualization environment superior to noninteractive software-based implementations, but with a rendering speed that is comparable to traditional, but inferior quality, volume rendering approaches based on texture mapping. This retrospective ECG-gated dynamic cardiac display system can provide real-time feedback regarding the suspected pathology, function, and structural defects, as well as anatomical information such as chamber volume and morphology.     

容积数据高分辨力CT重建诊断肺部磨玻璃密度影

目的：探讨容积数据高分辨力CT重建（VHRCT）对肺部磨玻璃密度影的临床诊断价值。方法：对27例肺部弥漫性磨玻璃密度灶和5例局限性磨玻璃密度灶患者进行MSCT和HRCT扫描及VHRCT图像重建,对HRCT及VHRCT图像质量进行分级评分（3级评分法）。结果：27例肺部弥漫性磨玻璃密度灶患者的VHRCT和HRCT图像评分结果分别是52分和49分,2种图像质量的差异无统计学意义（Z=-1.00,P=0.317）;5例局限性磨玻璃密度患者的VHRCT和HRCT图像评分结果分别1分和0分,2种图像质量的差异无统计学意义（Z=-1.00,P=0.317）。结论：VHRCT重建图像评价肺部磨玻璃密度影的价值与HRCT扫描图像相当,VHRCT重建图像为观察肺部磨玻璃密度影提供了一种简单易行的方法。     

Evaluation of the middle and inner ear structures: comparison of hybrid rendering, virtual endoscopy and axial 2D source images

Recent developments in 3D reconstructions can enhance the quality and diagnostic value of axial 2D image data sets with direct benefits for clinical practice. To show the possible advantages of a hybrid rendering method [color-coded 3D shaded-surface display (SSD)- and volume rendering method] with the possibility of virtual endoscopy we have specifically highlighted the use in relation to the middle and inner ear structures. We examined 12 patients with both normal findings and postoperative changes, using image data sets from high-resolution spiral computed tomography (HRSCT). The middle and inner ear was segmented using an interactive threshold interval density volume-growing method and visualized with a color-coded SSD rendering method. The temporal bone was visualized using a transparent volume rendering method. The 3D- and virtual reconstructions were compared with the axial 2D source images. The evaluated middle and inner ear structures could be seen in their complete form and correct topographical relationship, and the 3D- and virtual reconstructions indicated an improved representation and spatial orientation of these structures. A hybrid and virtual endoscopic method could add information and improve the value of imaging in the diagnosis and management of patients with middle or inner ear diseases making the understanding and interpretation of axial 2D CT image data sets easier. The introduction of an improved rendering algorithm aids radiological diagnostics, medical education, surgical planning, surgical training, and postoperative assessment. Received: 22 July 1998; Revised: 15 April 1999; Accepted: 19 April 1999     

Three-dimensional volume rendering of spiral CT data: theory and method.

Three-dimensional (3D) medical images of computed tomographic (CT) data sets can be generated with a variety of computer algorithms. The three most commonly used techniques are shaded surface display, maximum intensity projection, and, more recently, 3D volume rendering. Implementation of 3D volume rendering involves volume data management, which relates to operations including acquisition, resampling, and editing of the data set; rendering parameters including window width and level, opacity, brightness, and percentage classification; and image display, which comprises techniques such as "fly-through" and "fly-around," multiple-view display, obscured structure and shading depth cues, and kinetic and stereo depth cues. An understanding of both the theory and method of 3D volume rendering is essential for accurate evaluation of the resulting images. Three-dimensional volume rendering is useful in a wide variety of applications but is just now being incorporated into commercially available software packages for medical imaging. Although further research is needed to determine the efficacy of 3D volume rendering in clinical applications, with wider availability and improved cost-to-performance ratios in computing, 3D volume rendering is likely to enjoy widespread acceptance in the medical community.     

Use of 2D histograms for volume rendering of multidetector CT data: development of a graphical user interface

RATIONALE AND OBJECTIVES: Direct volume rendering reveals 3D information on anatomic structures without preprocessing the data. This increases the interest in this technique as a diagnostic tool. A fast and simple method for setting transfer functions is crucial for clinical routine work. However, this is still a complex task. Present commercial workstations are usually limited to design galleries and window/level functionality. MATERIALS AND METHODS: We present a graphical user interface for volume rendering of multidetector row CT data that permits a much more flexible specification of rendering parameters. A 2D histogram of CT density versus gradient magnitude facilitates the understanding of the spatial connections of different tissues. The incorporation of gradient magnitude into the transfer function domain allows discrimination of features of interest that are not distinguishable on CT density alone. Penetration length, color, and gradient magnitude are depicted on a stack of 2D slices according to the settings of the opacity transfer function and the viewing direction. A gallery of thumbnails with presets of transfer functions is interactively adapted if the volume is rotated or cropped. RESULTS: This allows for fast evaluation of numerous rendering protocols at once. The interface was evaluated with CT data covering skeletal trauma, pathologies of the thorax/abdomen, and CT angiography. CONCLUSION: We observed that high-quality visualizations could be obtained with reasonable interaction times. The 2D histogram and penetration length displays provided valuable insight into the dataset that made the specification of transfer functions a goal-oriented process.     

Local and remote visualization techniques for interactive direct volume rendering in neuroradiology.

The increasing capabilities of magnetic resonance (MR) imaging and multisection spiral computed tomography (CT) to acquire volumetric data with near-isotropic voxels make three-dimensional (3D) postprocessing a necessity, especially in studies of complex structures like intracranial vessels. Since most modern CT and MR imagers provide limited postprocessing capabilities, 3D visualization with interactive direct volume rendering requires expensive graphics workstations that are not available at many institutions. An approach has been developed that combines fast visualization on a low-cost PC system with high-quality visualization on a high-end graphics workstation that is directly accessed and remotely controlled from the PC environment via the Internet by using a Java client. For comparison of quality, both techniques were applied to several neuroradiologic studies: visualization of structures related to the inner ear, intracranial aneurysms, and the brainstem and surrounding neurovascular structures. The results of pure PC-based visualization were comparable with those of many commercially available volume-rendering systems. In addition, the high-end graphics workstation with 3D texture-mapping capabilities provides visualization results of the highest quality. Combining local and remote 3D visualization allows even small radiologic institutions to achieve low-cost but high-quality 3D visualization of volumetric data.     

Effects of the volume and shape of voxels on the measurement of phantom volume using three-dimensional magnetic resonance imaging

Recently, an increasing number of volumetric studies of the human brain have been reported, using three-dimensional magnetic resonance imaging (3D-MRI). To our knowledge, however, there are few investigations on the relation of the volume and shape of voxels which constitute an MR image to the accuracy in volume measurement of an imaged object. The purpose of this study was to evaluate the effect of a different shape of voxel, that is, isotropic or anisotropic, as well as the volume of a voxel on the volume measurement based on the original image data and multiplanar reconstruction (MPR) data, respectively. In the experiment, we repeatedly acquired contiguous sagittal images of a single globe phantom with a known volume under the condition in which the volume and shape of voxels varied, on a 1.5T MR scanner. We used a gradient echo sequence (3D FLASH). The volume of the globe phantom from both original images and MPR ones was measured on workstations employing a semi-automated local thresholding technique. As a result, the smaller volume of voxels tended to give us the more correct measurement, and an isotropic voxel reduced measurement errors as compared to an anisotropic one. Therefore, it is concluded that the setting of voxel with both an isotropic shape and small volume, e.g., a voxel of 1 mm x 1 mm x 1 mm at present, is recommended in order to get a precise volume measurement using 3D-MRI.