HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) CME Test (opens in a new window) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Fishman, E. K. Articles by Johnson, P. T. Search for Related Content PubMed PubMed Citation Articles by Fishman, E. K. Articles by Johnson, P. T. Related Collections Computed Tomography General Physics and Basic Science

Volume Rendering versus Maximum Intensity Projection in CT Angiography: What Works Best, When, and Why¹

¹ From the Russell H. Morgan Department of Radiology, Johns Hopkins School of Medicine, 601 N Caroline St, Room 3251, Baltimore, MD 21287 (E.K.F., D.G.H., F.M.C., K.M.H., P.T.J.); and HipGraphics, Towson, Md (D.R.N., D.G.H.). Presented as an education exhibit at the 2004 RSNA Annual Meeting. Received October 11, 2005; revision requested November 4 and received December 19; accepted December 20. D.G.H. and D.R.N. are founders of HipGraphics; D.G.H. is a consultant to HipGraphics; D.R.N. is a full-time employee of HipGraphics; E.K.F. is a co-founder of Hip-Graphics and a consultant to Siemens Medical Solutions and GE Healthcare; and the other authors have no financial relationships to disclose. Address correspondence to E.K.F. (e-mail: efishman{at}jhmi.edu).

	Abstract

Top Abstract LEARNING OBJECTIVES Introduction Rendering Techniques: Principles... Are All Images the... MIP versus Volume Rendering:... Preferential Uses and Pitfalls Conclusions References

 
The introduction and widespread availability of 16-section multi–detector row computed tomographic (CT) technology and, more recently, 64-section scanners, has greatly advanced the role of CT angiography in clinical practice. CT angiography has become a key component of state-of-the-art imaging, with applications ranging from oncology (eg, staging of pancreatic or renal cancer) to classic vascular imaging (eg, evaluation of aortic aneurysms and renal artery stenoses) as well as newer techniques such as coronary artery imaging and peripheral runoff studies. With an average of 400–1000 images in each volume data set, three-dimensional postprocessing is crucial to volume visualization. Radiologists now have workstations that provide capabilities for evaluation of these data sets by using a range of software programs and processing tools. Although different systems have unique capabilities and functionality, all provide the options of volume rendering and maximum intensity projection for image display and analysis. These two postprocessing techniques have different advantages and disadvantages when used in clinical practice, and it is important that radiologists understand when and how each technique should be used.

© RSNA, 2006

	LEARNING OBJECTIVES

Top Abstract LEARNING OBJECTIVES Introduction Rendering Techniques: Principles... Are All Images the... MIP versus Volume Rendering:... Preferential Uses and Pitfalls Conclusions References

 
After reading this article and taking the test, the reader will be able to:

• Describe the technical advantages and disadvantages of volume rendering and maximum intensity projection (MIP) for evaluation of CT angiographic data sets.
  
• Explain the value of volume rendering for optimizing tumor staging in solid organs such as the liver, pancreas, and kidney.
  
• Discuss the advantages of MIP for defining complex vascular anatomy.
  

	Introduction

Top Abstract LEARNING OBJECTIVES Introduction Rendering Techniques: Principles... Are All Images the... MIP versus Volume Rendering:... Preferential Uses and Pitfalls Conclusions References

 
Until recently, computed tomographic (CT) angiography was a special examination, performed for limited clinical indications at select institutions. However, the introduction and widespread availability of 16-section multi–detector row CT technology and, more recently, 64-section scanners, has greatly advanced the role of CT angiography in clinical practice. While progressive advances in multisection CT technology have dramatically enhanced the quality of three-dimensional (3D) renderings, the implementation of 64-section technology has resulted in a further improvement as well as an expansion of clinical applications. Sixty-four-section CT has higher temporal and spatial resolution than does 16-section CT, and scanning is three to four times faster. The increased speed facilitates the coupling of image acquisition with peak vascular enhancement and yields data sets without motion-related artifacts. In addition, cardiac gating is facilitated by 64-section CT technology.

CT angiography has been incorporated into mainstream radiology practice and is performed daily for a wide range of clinical indications. The success of CT angiography depends on a number of critical steps, including properly timed delivery of iodinated contrast material, correct timing of data acquisition, and selection of appropriate scanning parameters. The reconstruction of CT angiographic data sets obtained on 16- and 64-section scanners may result in 1000–5000 images per examination. The large size of the data set makes it impractical to extract all the information present by using standard two-dimensional techniques and makes clear the importance of volume imaging and 3D image display.

Traditionally, a radiologist would review a case in the axial plane. Next, he or she might review multiplanar reconstructions, and then, in selected cases, true 3D images. However, today, comprehensive review of a case requires a more integrated approach that is commonly referred to as volume visualization. The entire case can be approached as a volume of information to be reviewed as appropriate. This paradigm shift has required significant improvements in workstation design and software. Systems must process data quickly and be easy to use. A variety of vendors have been committed to improving 3D workstations, and, although systems do vary, most rely on a combination of 3D applications that includes volume rendering and maximum intensity projection (MIP). Although both techniques have been available on some systems for nearly two decades, they have undergone substantial evolution, especially in the past few years.

In this article, we describe the reconstruction of 3D images by using both volume rendering and MIP techniques and discuss the specific advantages, disadvantages, and potential pitfalls of each technique. Through case studies and illustrations, we also offer suggestions for optimizing the use of these techniques in daily practice.

	Rendering Techniques: Principles and Concepts

Top Abstract LEARNING OBJECTIVES Introduction Rendering Techniques: Principles... Are All Images the... MIP versus Volume Rendering:... Preferential Uses and Pitfalls Conclusions References

 
The rendering technique is an important technical determinant of 3D image quality in most circumstances. The rendering technique is the computer algorithm used to transform serially acquired axial CT image data into 3D images. There are a number of different methods, but most can be divided into the following two classes: thresholding- or surface-based (binary) techniques; and percentage- or semitransparent volume–based (continuous) techniques (1–3). The initial selection of the rendering technique greatly affects the quality of the reconstructed images in any 3D application.

Rendering with either technique consists of three steps: volume formation, classification, and image projection. Volume formation involves the acquisition of the image data, the stacking of the resultant data to form a volume, and preprocessing, which varies according to the rendering technique used. Typical preprocessing includes resizing (by interpolation or resampling) of each volume element (voxel), image smoothing, and data editing (eg, deletion of the CT table on which the patient was positioned). The classification step consists of determining the types of tissue (or other classifying quality) that are represented in each voxel and using that information to assign color and other visual properties to the voxel. The assigned values of visual characteristics can be either binary or continuous. At contrast material–enhanced CT, most voxels can be classified as a combination of two or more of the following basic tissue types: fat, soft tissue, bone, contrast-enhanced tissue, and air. Other imaging modalities may yield different categories of classification. The final step consists of projecting the classified volume data so that an image is displayed that represents a view of the 3D volume in the user-selected orientation.

Most early 3D imaging involved the use of thresholding-based techniques, since thresholding can easily produce a model of surfaces of objects within a volume, even with limited computer power. For thresholding (which is a binary, not a continuous classification technique), each type of tissue to be classified is assigned two numbers: a low and a high threshold. For a voxel to be considered to represent that tissue, its attenuation must lie within the range defined by the low and high thresholds (2). Bone is usually assigned a low threshold around 100 HU and a high threshold of more than 3000 HU (essentially, the top of the scale for most CT data sets).

In the classification of a volume, the attenuation value of each voxel is analyzed and compared with the low and high thresholds for each tissue. If the attenuation value is between the high and low thresholds defined for a tissue, the voxel is considered to contain that type of tissue. If the attenuation value lies outside the defined thresholds, the voxel is considered not to contain that tissue type. The defined ranges of thresholds for various tissue types should not overlap. This classification system is binary; that is, it defines each voxel as containing either 100% or 0% of a given tissue type, but nothing between those two percentages. Each tissue type is assigned a color (and possibly a level of transparency). Once the volume has been classified, most thresholding-based algorithms extract surfaces from the classified data. A surface is defined as a boundary between voxels of one tissue type and those of another tissue type. An image then can be generated by defining the viewing orientation, calculating which surfaces would be visible from such an orientation, and projecting the information into a two-dimensional viewing plane. The display may be reflective, with a simulated light source, or self-luminous. Both display types provide perspective and depth cues.

The thresholding-based technique of classification has a number of limitations (4,5), its greatest disadvantage being that voxels that represent mixed tissue interfaces cannot be accurately classified (1,6). Volume averaging occurs when two or more types of tissue are depicted within one voxel. Thus, in CT, a voxel that encompasses the boundary of muscle and bone contains a volume average of attenuation values for bone and soft tissue. All imaging modalities produce voxels with volume averaging, because voxels have a finite size; however, thresholding is based on the assumption that each volume element represents only one type of tissue. Thresholding-based classification is thus incompatible with volume averaging and incorrectly classifies voxels that contain volume averages. The effect of volume averaging is most significant at tissue-surface interfaces. Of the voxels along a periosteal surface, for instance, many represent both bone and soft tissue. This geometric reality makes the accurate depiction of surfaces difficult with the use of thresholding-based classification. Ubiquitous volume averaging makes it difficult to define a set of thresholds that represent a particular surface as it is modified by anatomic variation and pathologic conditions. The choice of a fixed threshold severely constrains this technique. The threshold that would approximate the attenuation of bone in a healthy patient, for instance, exceeds the attenuation value for markedly osteopenic bone, a situation that may create gaps in the CT data and inaccuracies in the reconstructed image. Similarly, in CT angiography, incorrect threshold selection may result in inaccurate grading of vascular stenosis (7–9). The thresholding technique is also susceptible to noise introduced during scanning. A small amount of noise can modify attenuation values and create the appearance of soft tissue in a voxel that actually represents mostly bone.

All of these disadvantages add up to a number of deleterious artifacts on the reconstructed image: holes in structures, contours that represent voxel boundaries rather than true tissue interfaces, fragments of structures floating in space, and absence or exaggeration of details such as bone fractures (4,5). The main advantage of thresholding-based reconstruction is its speed, since a comparatively small amount of computational power is needed to generate images in a reasonable amount of time. We do not use thresholding-based reconstruction techniques in our clinical practice.

Volume rendering, an alternative technique for 3D display of medical imaging data, came into use in the late 1980s. Volume rendering has an advantage over thresholding because it can be used to display data without classifying it into rigid all-or-nothing categories. Volume rendering is usually combined with percentage classification. The key difference between thresholding-based classification and percentage classification is that in the former it is assumed that each voxel contains either all or none of a particular tissue type, instead of a combination of different tissues. In percentage classification, it is assumed that a voxel may represent one or more tissue types and that the amount of each tissue, as a percentage of the entire voxel, is between 0% and 100%. Thus, percentage classification can more closely approximate the actual contents of voxels that represent various tissues or volume averages. Percentage classification involves the examination of each voxel to determine the amount of each type of tissue represented in the voxel. The resultant classified volume data consist of voxels in which each tissue represented is accounted for as a percentage of the whole voxel.

In the most common method used to calculate the tissue percentages for each voxel, a trapezoid or ramp is used for each tissue type. For example, let us consider the method used to calculate the percentage of bone. For bone, a simple ramp that is similar to a window width–window level curve can be used to obtain the percentages. A window width of 600 HU and a level of 400 HU closely approximate the percentages of bone. Voxels that appear dark on images obtained with that window width and window level are assigned 0% bone, and voxels that appear bright are assigned 100% bone. Voxels within the range between those two extremes are various shades of gray and contain a percentage of bone between 0% and 100%. For instance, with a window width of 600 HU and a level of 400 HU (which is a ramp that begins at 100 HU and increases to full brightness at 700 HU), a voxel with an attenuation value of less than 100 HU would be dark and therefore assigned 0% bone; a voxel with an attenuation value of 250 HU would be gray and would be assigned (250 HU–100 HU)/600 HU = 25% bone; and a voxel with an attenuation value greater than or equal to 700 HU would be assigned 100% bone.

For tissues such as soft tissue, an open-ended ramp does not work well because it cannot model the narrow range of attenuation values. In that case, a trapezoidal shape (which can be thought of as two ramps placed back-to-back, one of which slopes upward and the other of which slopes downward) is a good method for assigning percentages. As one moves up the scale of attenuation values, there are four points defined by the trapezoid: a low 0% point, a low 100% point, a high 100% point, and a high 0% point. (Again, to use the back-to-back ramp analogy, the low 0% to low 100% section is exactly the same as a window width–window level ramp, and the high 100% to high 0% section is a reverse ramp such as might be used for bright-air imaging.) An attenuation value between the low 0% point and the low 100% point is assigned a soft-tissue percentage within the range of 0%–100%. An attenuation value between the low 100% point and the high 100% point is assigned a soft-tissue percentage of exactly 100%. An attenuation value between the high 100% point and the high 0% point is assigned a soft-tissue percentage within the range of 100%–0% soft tissue. An attenuation value below the 0% low point or above the 0% high point is assigned 0% soft tissue. Typical values for these four points for soft tissue would be a low 0% point of 25 HU, a low 100% point of 50 HU, a high 100% point of 75 HU, and a high 0% point of 125 HU. Note that the ranges between different tissue types generally overlap: For example, the tail end of the downward ramp for soft tissue (100–125 HU) overlaps with the typical bone ramp, which begins its upward climb at 100 HU. The region of overlap is the region in which the voxels contain volume averages (ie, both bone and soft tissue), and percentage classification can represent this accurately.

Once the data have been assigned percentages, they must be further processed to form a final image. Each tissue type is given a color and transparency, and thus each voxel can be assigned a color and transparency by calculating the weighted sum of the percentage of each tissue type represented in the voxel and the color and transparency assigned to that tissue type. For example, assume that soft tissue is assigned the color red with a 50% transparency and that bone is assigned the color white with a 25% transparency. A voxel that represented 75% soft tissue and 25% bone would be a pinkish color and have a transparency between 50% and 25%. Once color and transparency are assigned to each voxel, a 3D image is produced by casting simulated rays of light through the volume that contains the classified and colored voxels (Fig 1). As the simulated rays pass through a voxel, the color and transparency of the voxel modulate the color of the rays. The final result is an image that can be displayed on a computer monitor or as hard copy. Volume rendering requires more computational power than do surface-based reconstruction techniques because each voxel in the data set must be projected into an image; with a surface-based technique, only the surface data must be processed. Images generated with the use of volume rendering do not have many of the significant computer-generated artifacts found on images reconstructed with surface-based or threshold-based classification (10). Computer-generated artifacts, at best, tend to engender distrust of 3D images and, at worst, could lead to profound diagnostic or therapeutic errors. We believe the greater fidelity of volume rendering to the patient data justifies the additional computational power required.

View larger version (57K): [in this window] [in a new window] [Download PPT slide]   Figure 1a.  Schematic comparison of volume-rendered and MIP images. (a) Volume-rendered image provides clear definition of individual vessels. (b) MIP image reconstructed from the same volume data shows all of the vessels, but their outlines merge; it is impossible to visualize the spatial relationships between the vessels or to delineate individual vessels on the MIP image.

View larger version (56K): [in this window] [in a new window] [Download PPT slide]   Figure 1b.  Schematic comparison of volume-rendered and MIP images. (a) Volume-rendered image provides clear definition of individual vessels. (b) MIP image reconstructed from the same volume data shows all of the vessels, but their outlines merge; it is impossible to visualize the spatial relationships between the vessels or to delineate individual vessels on the MIP image.

If thresholding and volume rendering are considered the two end-points of a line, then MIP lies somewhere between them. In the MIP algorithm, the highest-attenuation voxels along lines projected through the volume data set are selected (3). The subset of these high-attenuation voxels from the volume is then incorporated into a two-dimensional image. MIP is a widely used rendering tool and has proved particularly valuable for evaluation and display in CT angiography (21–26). MIP has been shown to be more accurate than surface rendering for evaluating the vasculature (8,27,28). However, an understanding of how the MIP algorithm produces renderings and of the limitations of MIP is essential for the correct interpretation of MIP images. The two most significant limitations of the technique are that the presence of other high-attenuation voxels may obscure evaluation of the vasculature and that the 3D relationships among the structures in the display are not visible (2). Most commonly, this first limitation is a problem when evaluating atherosclerotic vessels that contain large calcifications, because the calcifications prevent accurate visualization of the vascular lumen (12,29,30). The second limitation results from the fact that the MIP display is a two-dimensional representation that cannot accurately depict the actual 3D relationships of the vessels (29,31,32) (Fig 1). To improve evaluation of the vasculature by removing other high-attenuation voxels either from adjacent vessels or from bone, slab editing of the volume is helpful (Figs 2, 3) (18,31). The advantage of volume editing was confirmed in a recent renal donor study in which sliding thin-slab reconstruction was compared with thick-slab reconstruction with use of both volume rendering and MIP. Results showed that sliding thin-slab reconstruction significantly increased the sensitivity of CT for the detection of supernumerary arteries (18). Despite the shortcomings of this algorithm, MIP has become a valuable tool for 3D rendering of the vasculature.

View larger version (52K): [in this window] [in a new window] [Download PPT slide]   Figure 2a.  Schematic comparison of volume-rendered and MIP images after editing of the data set. (a) The large volume is reduced to a thinner slab (15 mm) by using an automated clip plane editing tool. (b) With volume rendering, vessels are well defined and the 3D spatial relationships between individual vessels can be correctly delineated. (c) With MIP, each vessel is defined but there is a lack of separation and accurate spatial orientation.

View larger version (62K): [in this window] [in a new window] [Download PPT slide]   Figure 2b.  Schematic comparison of volume-rendered and MIP images after editing of the data set. (a) The large volume is reduced to a thinner slab (15 mm) by using an automated clip plane editing tool. (b) With volume rendering, vessels are well defined and the 3D spatial relationships between individual vessels can be correctly delineated. (c) With MIP, each vessel is defined but there is a lack of separation and accurate spatial orientation.

View larger version (61K): [in this window] [in a new window] [Download PPT slide]   Figure 2c.  Schematic comparison of volume-rendered and MIP images after editing of the data set. (a) The large volume is reduced to a thinner slab (15 mm) by using an automated clip plane editing tool. (b) With volume rendering, vessels are well defined and the 3D spatial relationships between individual vessels can be correctly delineated. (c) With MIP, each vessel is defined but there is a lack of separation and accurate spatial orientation.

View larger version (55K): [in this window] [in a new window] [Download PPT slide]   Figure 3a.  Schematic comparison of volume-rendered and MIP images after further editing of the data set. (a) The slab is narrowed further (to 5 mm) with clip plane editing of the volume. (b, c) Volume-rendered (b) and MIP (c) images both depict the individual vessels. However, despite the thinner slab, the image in c is a flat projection and does not provide a 3D view of the vessels.

View larger version (63K): [in this window] [in a new window] [Download PPT slide]   Figure 3b.  Schematic comparison of volume-rendered and MIP images after further editing of the data set. (a) The slab is narrowed further (to 5 mm) with clip plane editing of the volume. (b, c) Volume-rendered (b) and MIP (c) images both depict the individual vessels. However, despite the thinner slab, the image in c is a flat projection and does not provide a 3D view of the vessels.

View larger version (60K): [in this window] [in a new window] [Download PPT slide]   Figure 3c.  Schematic comparison of volume-rendered and MIP images after further editing of the data set. (a) The slab is narrowed further (to 5 mm) with clip plane editing of the volume. (b, c) Volume-rendered (b) and MIP (c) images both depict the individual vessels. However, despite the thinner slab, the image in c is a flat projection and does not provide a 3D view of the vessels.

	Are All Images the Same?

Top Abstract LEARNING OBJECTIVES Introduction Rendering Techniques: Principles... Are All Images the... MIP versus Volume Rendering:... Preferential Uses and Pitfalls Conclusions References

In addition, in most volume rendering methods there are many adjustable parameters that change the way the image looks. The simplest parameters are window settings (window center and width). Other volume rendering parameters include color, degree of opacity or transparency, and shading. Because there is no standardization of volume rendering methods among vendors, the parameters from one system do not generally translate well to another.

Image quality also varies among vendors. The particular volume rendering method used has the greatest effect on image quality, but other factors also come into play. Does the system use full 12-bit (–1024 to 3072 HU) input data for rendering? Does it limit the volume size to some maximum, so that larger volumes are shrunk when loaded? What is the quality of the video hardware display? All these factors affect image quality.

MIP is a specific type of rendering in which the brightest voxel is projected into the 3D image. There tends to be much less variability in MIP image reconstruction than in volume rendering, because fewer parameters are factored in to the MIP algorithm. Most MIP methods use only windowing parameters (window width and center, specified in Hounsfield units) and not color, opacity, or shading; windowing parameters are standardized measures that should be valid across different vendors’ systems. Nonetheless, it has been reported that differences in image quality result from reconstruction with different MIP algorithms (33). Other image quality factors (12-bit input data, volume reduction, and display) besides windowing may affect MIP reconstruction. For large cases (eg, runoff cases with more than 1000 sections), volume reduction can have a huge effect on MIP image quality, and the use of the entire range of input data (12 bits) can have a lesser effect.

	MIP versus Volume Rendering: Imaging Examples

Top Abstract LEARNING OBJECTIVES Introduction Rendering Techniques: Principles... Are All Images the... MIP versus Volume Rendering:... Preferential Uses and Pitfalls Conclusions References

 
Volume rendering always accurately depicts 3D relationships, while MIP may not do so, especially on arterial phase–dominant images that show both arterial and venous structures (Figs 4–6). On such MIP images, venous structures typically appear to have an anatomic location that is more posterior than their actual location. Volume rendering not only allows display of the vascular anatomy but also provides definition of soft tissue, muscle, and bone, which may contribute to a more comprehensive understanding of pathologic processes (Figs 7a, 7b, 8). In addition, volume rendering enables a color display, which often improves the visualization of complex anatomy and 3D relationships (32) (Figs 9–11). However, MIP may allow visualization of smaller branch vessels with less work than is required for volume rendering. The capability of viewing only the brightest pixels may help to define smaller branch vessels in the liver (21), kidney, and lung (Figs 11, 12).

View larger version (119K): [in this window] [in a new window] [Download PPT slide]   Figure 4a.  Renal donor evaluation. (a) Coronal oblique volume-rendered image depicts classic orientation of the renal vein, two left renal arteries, and a left prehilar renal artery branch (small arrow). The left gonadal vein (large arrow) is well defined. (b) Coronal oblique MIP image correctly defines the renal arteries, but the locations of the renal vein and gonadal vein (arrow) are inaccurately depicted.

View larger version (143K): [in this window] [in a new window] [Download PPT slide]   Figure 4b.  Renal donor evaluation. (a) Coronal oblique volume-rendered image depicts classic orientation of the renal vein, two left renal arteries, and a left prehilar renal artery branch (small arrow). The left gonadal vein (large arrow) is well defined. (b) Coronal oblique MIP image correctly defines the renal arteries, but the locations of the renal vein and gonadal vein (arrow) are inaccurately depicted.

View larger version (98K): [in this window] [in a new window] [Download PPT slide]   Figure 5a.  Liver mass evaluation. (a) Coronal oblique volume-rendered image provides good 3D definition of the superior mesenteric artery, the celiac artery, and the tortuous splenic artery and hepatic artery. (b) On the coronal oblique MIP image, the 3D relationships are lost because of the rendering technique.

View larger version (147K): [in this window] [in a new window] [Download PPT slide]   Figure 5b.  Liver mass evaluation. (a) Coronal oblique volume-rendered image provides good 3D definition of the superior mesenteric artery, the celiac artery, and the tortuous splenic artery and hepatic artery. (b) On the coronal oblique MIP image, the 3D relationships are lost because of the rendering technique.

View larger version (129K): [in this window] [in a new window] [Download PPT slide]   Figure 6a.  Liver mass evaluation. (a) Coronal volume-rendered image depicts a replaced right hepatic artery (small arrow) that branches from the superior mesenteric artery, as well as two hepatic hemangiomas (large arrows). (b) Coronal MIP image depicts the hepatic lesions (large arrows) but inaccurately indicates that the right hepatic artery (small arrow) arises from the right renal artery.

View larger version (127K): [in this window] [in a new window] [Download PPT slide]   Figure 6b.  Liver mass evaluation. (a) Coronal volume-rendered image depicts a replaced right hepatic artery (small arrow) that branches from the superior mesenteric artery, as well as two hepatic hemangiomas (large arrows). (b) Coronal MIP image depicts the hepatic lesions (large arrows) but inaccurately indicates that the right hepatic artery (small arrow) arises from the right renal artery.

View larger version (211K): [in this window] [in a new window] [Download PPT slide]   Figure 7a.  Occlusion of the superior vena cava. (a, b) Volume-rendered images obtained with different rendering parameters demonstrate the flexibility of volume rendering for visualization of the chest wall musculature, thoracic bone, and the superior vena cava, which is nearly occluded by a tumor. (c) MIP image obtained with thin-slab reconstruction also shows near-occlusion of the superior vena cava.

View larger version (187K): [in this window] [in a new window] [Download PPT slide]   Figure 7b.  Occlusion of the superior vena cava. (a, b) Volume-rendered images obtained with different rendering parameters demonstrate the flexibility of volume rendering for visualization of the chest wall musculature, thoracic bone, and the superior vena cava, which is nearly occluded by a tumor. (c) MIP image obtained with thin-slab reconstruction also shows near-occlusion of the superior vena cava.

View larger version (132K): [in this window] [in a new window] [Download PPT slide]   Figure 7c.  Occlusion of the superior vena cava. (a, b) Volume-rendered images obtained with different rendering parameters demonstrate the flexibility of volume rendering for visualization of the chest wall musculature, thoracic bone, and the superior vena cava, which is nearly occluded by a tumor. (c) MIP image obtained with thin-slab reconstruction also shows near-occlusion of the superior vena cava.

View larger version (72K): [in this window] [in a new window] [Download PPT slide]   Figure 8a.  CT images obtained to rule out abscess. Coronal volume-rendered images of the hand and wrist from a posterior orientation (a with different reconstruction parameters than b) show soft-tissue swelling and localized erythema due to cellulitis, but no abscess.

View larger version (75K): [in this window] [in a new window] [Download PPT slide]   Figure 8b.  CT images obtained to rule out abscess. Coronal volume-rendered images of the hand and wrist from a posterior orientation (a with different reconstruction parameters than b) show soft-tissue swelling and localized erythema due to cellulitis, but no abscess.

View larger version (74K): [in this window] [in a new window] [Download PPT slide]   Figure 9a.  CT angiography depicts an enlarged mass at a prior anastomotic site in the right side of the groin. Coronal (a) and coronal oblique (b) color-coded volume-rendered images provide realistic 3D views of the pseudoaneurysm, feeding and draining vessels, and occlusion of the left limb of the graft.

View larger version (101K): [in this window] [in a new window] [Download PPT slide]   Figure 9b.  CT angiography depicts an enlarged mass at a prior anastomotic site in the right side of the groin. Coronal (a) and coronal oblique (b) color-coded volume-rendered images provide realistic 3D views of the pseudoaneurysm, feeding and draining vessels, and occlusion of the left limb of the graft.

View larger version (114K): [in this window] [in a new window] [Download PPT slide]   Figure 10a.  Evaluation of an endovascular stent with CT angiography. Coronal (a, c) and sagittal (b, d) MIP images (a, b) and volume-rendered images (c, d) show successful stent placement. The sagittal image with color mapping in d provides improved depiction of the stent detail, as well as a more accurate 3D perspective.

View larger version (95K): [in this window] [in a new window] [Download PPT slide]   Figure 10b.  Evaluation of an endovascular stent with CT angiography. Coronal (a, c) and sagittal (b, d) MIP images (a, b) and volume-rendered images (c, d) show successful stent placement. The sagittal image with color mapping in d provides improved depiction of the stent detail, as well as a more accurate 3D perspective.

View larger version (81K): [in this window] [in a new window] [Download PPT slide]   Figure 10c.  Evaluation of an endovascular stent with CT angiography. Coronal (a, c) and sagittal (b, d) MIP images (a, b) and volume-rendered images (c, d) show successful stent placement. The sagittal image with color mapping in d provides improved depiction of the stent detail, as well as a more accurate 3D perspective.

View larger version (76K): [in this window] [in a new window] [Download PPT slide]   Figure 10d.  Evaluation of an endovascular stent with CT angiography. Coronal (a, c) and sagittal (b, d) MIP images (a, b) and volume-rendered images (c, d) show successful stent placement. The sagittal image with color mapping in d provides improved depiction of the stent detail, as well as a more accurate 3D perspective.

View larger version (181K): [in this window] [in a new window] [Download PPT slide]   Figure 11a.  CT angiography of the pulmonary vasculature. Coronal MIP images (a, b) and volume-rendered images (c, d) based on a volume data set obtained with 64-section multi–detector row CT show an amazing level of detail. Color mapping helps increase the 3D effect in d. The MIP images show a bit more vessel detail at the periphery, produced with less operator interaction than was detail on the volume-rendered images.

View larger version (194K): [in this window] [in a new window] [Download PPT slide]   Figure 11b.  CT angiography of the pulmonary vasculature. Coronal MIP images (a, b) and volume-rendered images (c, d) based on a volume data set obtained with 64-section multi–detector row CT show an amazing level of detail. Color mapping helps increase the 3D effect in d. The MIP images show a bit more vessel detail at the periphery, produced with less operator interaction than was detail on the volume-rendered images.

View larger version (181K): [in this window] [in a new window] [Download PPT slide]   Figure 11c.  CT angiography of the pulmonary vasculature. Coronal MIP images (a, b) and volume-rendered images (c, d) based on a volume data set obtained with 64-section multi–detector row CT show an amazing level of detail. Color mapping helps increase the 3D effect in d. The MIP images show a bit more vessel detail at the periphery, produced with less operator interaction than was detail on the volume-rendered images.

View larger version (206K): [in this window] [in a new window] [Download PPT slide]   Figure 11d.  CT angiography of the pulmonary vasculature. Coronal MIP images (a, b) and volume-rendered images (c, d) based on a volume data set obtained with 64-section multi–detector row CT show an amazing level of detail. Color mapping helps increase the 3D effect in d. The MIP images show a bit more vessel detail at the periphery, produced with less operator interaction than was detail on the volume-rendered images.

View larger version (117K): [in this window] [in a new window] [Download PPT slide]   Figure 12a.  CT angiography for evaluation of renal artery stenosis. Both the volume-rendered image (a) and the MIP image (b) provide excellent vessel depiction in the coronal plane, but b shows a bit more of the peripheral intrarenal vessels.

View larger version (132K): [in this window] [in a new window] [Download PPT slide]   Figure 12b.  CT angiography for evaluation of renal artery stenosis. Both the volume-rendered image (a) and the MIP image (b) provide excellent vessel depiction in the coronal plane, but b shows a bit more of the peripheral intrarenal vessels.

View larger version (88K): [in this window] [in a new window] [Download PPT slide]   Figure 13a.  CT angiography of the carotid artery. (a) Sagittal oblique MIP image suggests internal carotid artery occlusion. (b, c) Sagittal oblique (b) and sagittal (c) opaque 3D volume-rendered images show that the calcification does not cause luminal narrowing. Both opaque and transparent volume-rendered images are helpful in this application.

View larger version (97K): [in this window] [in a new window] [Download PPT slide]   Figure 13b.  CT angiography of the carotid artery. (a) Sagittal oblique MIP image suggests internal carotid artery occlusion. (b, c) Sagittal oblique (b) and sagittal (c) opaque 3D volume-rendered images show that the calcification does not cause luminal narrowing. Both opaque and transparent volume-rendered images are helpful in this application.

View larger version (100K): [in this window] [in a new window] [Download PPT slide]   Figure 13c.  CT angiography of the carotid artery. (a) Sagittal oblique MIP image suggests internal carotid artery occlusion. (b, c) Sagittal oblique (b) and sagittal (c) opaque 3D volume-rendered images show that the calcification does not cause luminal narrowing. Both opaque and transparent volume-rendered images are helpful in this application.

View larger version (63K): [in this window] [in a new window] [Download PPT slide]   Figure 14a.  CT angiography for evaluation of peripheral vascular disease. Extensive calcifications make it impossible to assess luminal patency on the posterior coronal MIP images (a, b). On the posterior coronal volume-rendered images (c, d), the calcifications are depicted on the vessel walls, and luminal patency is well defined.

View larger version (98K): [in this window] [in a new window] [Download PPT slide]   Figure 14b.  CT angiography for evaluation of peripheral vascular disease. Extensive calcifications make it impossible to assess luminal patency on the posterior coronal MIP images (a, b). On the posterior coronal volume-rendered images (c, d), the calcifications are depicted on the vessel walls, and luminal patency is well defined.

View larger version (75K): [in this window] [in a new window] [Download PPT slide]   Figure 14c.  CT angiography for evaluation of peripheral vascular disease. Extensive calcifications make it impossible to assess luminal patency on the posterior coronal MIP images (a, b). On the posterior coronal volume-rendered images (c, d), the calcifications are depicted on the vessel walls, and luminal patency is well defined.

View larger version (67K): [in this window] [in a new window] [Download PPT slide]   Figure 14d.  CT angiography for evaluation of peripheral vascular disease. Extensive calcifications make it impossible to assess luminal patency on the posterior coronal MIP images (a, b). On the posterior coronal volume-rendered images (c, d), the calcifications are depicted on the vessel walls, and luminal patency is well defined.

View larger version (86K): [in this window] [in a new window] [Download PPT slide]   Figure 15a.  CT angiography to rule out vascular injury due to trauma. (a, b) Anterior (a) and posterior (b) coronal volume-rendered images define the fracture as well as the patent popliteal artery. Editing is often unnecessary when opaque volume rendering is used, as in this case. (c, d) Anterior coronal MIP images require editing of bone to enable visualization of the vascular map.

View larger version (86K): [in this window] [in a new window] [Download PPT slide]   Figure 15b.  CT angiography to rule out vascular injury due to trauma. (a, b) Anterior (a) and posterior (b) coronal volume-rendered images define the fracture as well as the patent popliteal artery. Editing is often unnecessary when opaque volume rendering is used, as in this case. (c, d) Anterior coronal MIP images require editing of bone to enable visualization of the vascular map.

View larger version (70K): [in this window] [in a new window] [Download PPT slide]   Figure 15c.  CT angiography to rule out vascular injury due to trauma. (a, b) Anterior (a) and posterior (b) coronal volume-rendered images define the fracture as well as the patent popliteal artery. Editing is often unnecessary when opaque volume rendering is used, as in this case. (c, d) Anterior coronal MIP images require editing of bone to enable visualization of the vascular map.

View larger version (71K): [in this window] [in a new window] [Download PPT slide]   Figure 15d.  CT angiography to rule out vascular injury due to trauma. (a, b) Anterior (a) and posterior (b) coronal volume-rendered images define the fracture as well as the patent popliteal artery. Editing is often unnecessary when opaque volume rendering is used, as in this case. (c, d) Anterior coronal MIP images require editing of bone to enable visualization of the vascular map.

View larger version (126K): [in this window] [in a new window] [Download PPT slide]   Figure 16a.  CT angiography of the liver in a patient with hepatitis C. Coronal (a, b) and axial (c, d) volume-rendered images (a, c) and complementary MIP images (b, d) define the hepatoma in the right lobe of the liver and provide a hepatic arterial map.

View larger version (116K): [in this window] [in a new window] [Download PPT slide]   Figure 16b.  CT angiography of the liver in a patient with hepatitis C. Coronal (a, b) and axial (c, d) volume-rendered images (a, c) and complementary MIP images (b, d) define the hepatoma in the right lobe of the liver and provide a hepatic arterial map.

View larger version (120K): [in this window] [in a new window] [Download PPT slide]   Figure 16c.  CT angiography of the liver in a patient with hepatitis C. Coronal (a, b) and axial (c, d) volume-rendered images (a, c) and complementary MIP images (b, d) define the hepatoma in the right lobe of the liver and provide a hepatic arterial map.

View larger version (118K): [in this window] [in a new window] [Download PPT slide]   Figure 16d.  CT angiography of the liver in a patient with hepatitis C. Coronal (a, b) and axial (c, d) volume-rendered images (a, c) and complementary MIP images (b, d) define the hepatoma in the right lobe of the liver and provide a hepatic arterial map.

View larger version (150K): [in this window] [in a new window] [Download PPT slide]   Figure 17a.  CT angiography with use of both MIP and volume rendering as complementary techniques for vascular mapping. (a) Coronal MIP image provides good definition of the gastroduodenal artery (arrow). (b) Coronal volume-rendered image shows a pancreatic adenocarcinoma that encases the vessel. (c) Coronal MIP image shows right renal artery stenosis due to noncalcified plaque in the proximal artery. (d) Coronal volume-rendered image elucidates the lower-attenuation soft plaque.

View larger version (150K): [in this window]