Image Post-Processing and Interpretation

Image post-processing is defined as the process of integrating a series of axial images into a form that is often easier to interpret than the sections themselves. The source transverse section is a fundamental image for interpretation. However, because of the complexity of coronary anatomy, interpreters must review coronary CT angiography interactively on workstations capable of 3-dimensional (3D) displays. The 3D data set can be reformatted according to the needs of the investigator using several types of post-processing algorithms. Coronary artery interpretation includes assessment of calcium scoring on non-contrast image and coronary angiography on enhanced image. Non-coronary cardiovascular findings and extra-cardiac findings should be reviewed in addition to recognize primary and secondary comorbid pathology and to identify findings that lead to alternative non-cardiovascular diagnoses. The ultimate objective of post-processing and interpretation is to convey diagnostic information to the treating physician with as much clarity and accuracy as possible. Interpretation times vary, usually depending on the image quality and the user’s experience and knowledge of cardiac CT and the individual workstation. Normally, experienced users usually spend no more than 10 minutes segmenting and interpreting cardiac CT studies.


Introduction
Image post-processing is defined as the process of integrating a series of axial images into a form that is often easier to interpret than the sections themselves.The source transverse section is a fundamental image for interpretation.However, because of the complexity of coronary anatomy, interpreters must review coronary CT angiography interactively on workstations capable of 3-dimensional (3D) displays.The 3D data set can be reformatted according to the needs of the investigator using several types of post-processing algorithms.Coronary artery interpretation includes assessment of calcium scoring on non-contrast image and coronary angiography on enhanced image.Non-coronary cardiovascular findings and extra-cardiac findings should be reviewed in addition to recognize primary and secondary comorbid pathology and to identify findings that lead to alternative non-cardiovascular diagnoses.The ultimate objective of post-processing and interpretation is to convey diagnostic information to the treating physician with as much clarity and accuracy as possible.Interpretation times vary, usually depending on the image quality and the user's experience and knowledge of cardiac CT and the individual workstation.Normally, experienced users usually spend no more than 10 minutes segmenting and interpreting cardiac CT studies.

Post-processing technique
The source transverse sections are the primary tools for interpretation of coronary CT angiography examinations.Post-processing imaging is performed from reconstructed axial images and is useful in allowing clinically relevant information to be extracted from large number of axial images.In addition, the diagnostic accuracy of axial images (86% sensitive and 78% accurate) for the detection of coronary stenosis is low compared with combination of post-processing techniques (98% sensitive and 94% accurate) (Jinzaki et al, 2009).Thus, the review of 3D displays is necessary for the coronary artery interpretation due to the complexity of coronary anatomy.These include volume rendering, multiplanar reformation, thin-slab maximum intensity projection, curved multiplanar reformation, angiographic view, and plaque-loaded angiographic view.A post-processing image which enables the evaluation of coronary artery disease with fewer images and one that is understandable to the third person is preferable.

Multiplanar reformation (MPR)
MPR is a high-resolution reconstruction format that allows display of planar images at any angular section through the acquisition volume, which permits visualization in not only the www.intechopen.comAdvances in the Diagnosis of Coronary Atherosclerosis 306 axial plane but also in orthogonal (coronal and sagittal) or oblique planes that better follow the arterial course in the thorax.The slice thickness can be modified, and a thick slab may be helpful in visualizing tortuous coronary artery.MPR is a simple algorithm which can be rendered quickly and accurately on any image-processing workstations.The distance measurements in MPR are accurate.Different structures in MPR are accurate and do not overlap.There is no loss of voxel value information due to threshold.MPRs are best suited for evaluation of the individual cardiac chambers, aorta (Fig. 1), and pulmonary vasculature with cardiac CT.MPRs are also well suited for selective evaluation of a coronary short segment.If a narrowing is suspected on axial views, display in coronal or sagittal orientations can confirm the presence of a stenosis (Fig. 2).The disadvantage of MPR is that only short segments can be analyzed at a time due to the nonlinear pathway of the coronary arteries.Thus, MPR requires the creation of numerous images in different planes at every suspected location of coronary stenosis.Furthermore, it is less suited for displaying the length of an artery.Grading of a stenosis is not advisable with coronal, sagittal, or oblique reconstructions alone, because of partial averaging (Johnson et al, 2010).

307
When a narrowing (arrow) is suspected on axial views (a), multiplanar reformation can be used to set the planes (b, c) in an orientation that matches one of the coronary short segments.

Thin-slab maximum intensity projection (Thin-slab MIP)
In both thin-slab MIP and MPR, orthogonal or oblique planes can be reviewed interactively.
MIP is a projection technique, in which only the highest density pixel in the projection direction is displayed to reconstruct the two-dimensional images.MIP image allows the differentiation between enhanced vascular structures and non-vascular structures, because the result is a projection of the highest density.In practice, MIP is usually performed interactively with a sliding slab, enabling the radiologist to adjust the window, select the optimal orientation to display each artery, and modify the slab thickness (Fishman et al, 2006).
The advantage of thin-slab MIP is the creation of a vascular map quickly without modification of several parameters (Fig. 3).In addition, MIP is better for displaying smallcaliber segments: the minimum vessel diameter seen with MIP was 0.7 mm versus 1.5 mm for volume rendering (VR) (Zhang et al, 2004).This is because MIP is designed to selectively demonstrate the arterial vasculature, which involves selection of the highest attenuation voxels along rays projected through the volume.It is controversial whether thin-slab MIP can be used for the diagnosis of coronary artery stenosis.One study compared interpretation of 3 sets of orthogonally oriented thin-slab MIPs with axial sections with 16slice CT (Choi et al, 2006).For all three radiologists, the Az (area under the ROC curve) values from the MIP images were higher than those Az values from the transaxial images, however, the difference was significant for only one reader.MIP has a tendency to overestimate stenosis.Heavily calcified arteries cannot be evaluated with MIP.Thus, thin-slab MIP cannot be used for the quantity evaluation of plaque, including the determination of grading of a stenosis.In addition, numerous images in different planes are required for the evaluation of the whole coronary arteries.Thin-slab MIP is useful as an adjunct to VR.However, it should not be the sole technique used for interpretation.

Curved multiplanar reformation (curved MPR)
The development of curved MPR format made it possible for the interpreter to follow the course of a tortuous vessel for longer distances as it changes direction (Fig. 4) (Ferencik et al., 2007).The artery can be displayed as a straight line (Fig. 5).This method enables one entire coronary artery to be followed and displayed in one reconstruction, and reduces the number of planar images needed, compared with MPR.An automated program is possible to track the coronary arteries using placement of a single point in any portion of each artery or ascending aorta, the computer can trace a given coronary vessel departing proximally and distally from the point chosen, and extract the images in seconds.Other automated program tracks the path of the coronary artery, when the operator selects two points: one at the beginning and one at the end of the vessel.These applications also provide MPRs that are orthogonal to the vessel centerline, thus creating cross-sectional cuts of each coronary arterial segment (Fig. 5).All other surrounding structures are automatically eliminated that are any distance from this centerpoint, possibly decreasing problems seen with other reconstruction methods from tortuosity and overlap.Curved MPR with coronary CT angiography has been reported with high sensitivity and specificity (Achenbach et al., 1998).Significant stenosis and occlusions of the coronary arteries can be visualized with an accuracy that exceeds 90%, since the CT value is preserved, making it fairly easy to delineate calcifications from the lumen.Recognition of calcifications with either higher or lower density than inner lumen contrast is very difficult with VR, somewhat better with MIP, and best with MPR.In addition, curved MPR is used to identify and quantify the degree of stenosis without being hindered by partial averaging (Fig. 5).This method is also useful in delineating the morphology of plaque and its effect on the lumen and adjacent vessel wall.Furthermore, the distance measurements in the curved MPR images represent the true distance.Overall, curved MPR, because they can be rendered quickly and convey all the information about the different CT densities within the vessel, is an extremely useful tool to evaluate contrast-enhanced CT angiography data.
The drawback of curved MPR is that only one vessel can be visualized in one image and that side branches are not depicted unless separate reconstructions are rendered for every side branch (Achenbach et al., 1998).Angiographic view is better in evaluating multiple branches.Furthermore, accurate assessment for stenosis requires correct centerline placement.Since the quality and ability of the computer workstation for tracking the vessel depends on border definition and overall image quality, off-axis centerline placement may happen, which result in misinterpretation of stenosis.Thus, curved MPR requires some editing in most cases.This commonly occurs in the presence of heavy arterial calcification and at the site of origin of branch.

Volume rendering (VR)
Rendering is a term taken from computer graphics and represents the process of generating a 3D model from a 2D image.Opacity and color are assigned to each CT value interval via transfer function, which is selectable and interactively modifiable (Calhoun et al., 1999).For example, normal soft tissue is assigned high transparency, enhanced vascular structures slight opaqueness and bone strong opaqueness.The sum of all CT values along each search ray from the observer through the 3D data volume, weighted by the transfer function, is displayed.In addition, color is used in most cases, with the color intensity decreasing with increasing distance to the observer in order to generate depth information and a 3D impression.The image can be rotated to different views, allowing the interpreter some flexibility in seeing the segment from multiple angles (Fig. 6).
VR provides an excellent overview of the cardiac and vascular anatomy, such as defining the course of coronary anomalies (Fig. 7a), the presence and course of coronary bypass grafts (Fig. 7b), the analysis of thoracic cardiovascular anatomy and congenital heart disease (Fig. 7c).Because VR creates volumetric 3D representations with the illusion of spatial integrity, it is helpful for surgeons to understand the anatomic complexity before surgery, and for teaching purposes and as illustrations for patients.
It is controversial whether VR can be used for the assessment of coronary stenosis.One study using 16 slice CT compared 3D VRs with combination of techniques including axial images, MPRs, and curved MPRs and MIPs.Sensitivity (63% vs 74%), specificity (80% vs 76%), and accuracy (78% vs 75%) were not significantly different between VR and the combination of techniques, respectively (Cordeiro et al., 2006).However, the diameter of vessels and the apparent thickness of the vessel lumen are dependent on threshold or transparency settings and the computer algorithm that is used to subtract non-vascular structures with VR (Ferencik et al., 2007) (Fig. 8).Furthermore, the overlap of cardiac veins on coronary arteries often hinders the coronary artery stenosis at the overlapped segment on VR (Fig. 9).Another disadvantage of VR is that due to its use of threshold or transparency setting which obscure the lower attenuation structures, the visualization of smaller vessel may be inferior compared with other displaying methods (Johnson et al, 2010).VR also requires several angles to evaluate the entire coronary artery (Fig. 6).

Angiographic view (AGV)
AGV is an whole heart MIP image in which contrast media of the ventricles is eliminated (Fig. 10).This image is similar to that from invasive conventional coronary angiography (CAG).
In our institute, AGV images divided into right and left coronary artery were rendered interactively from at least three angles for right coronary artery and six angles for left coronary artery (Fig. 11).If necessary, reconstructions were rendered from additional angles to visualize the entire coronary tree.AGV was originally automatically created only in GE workstation, but is available in many other workstations.It can be created in 3 steps after VR of cardiac CT is created in the usual method (Jinzaki et al., 2006).First, the origins of the right and left coronary arteries are separated from the ascending aorta.Second, the image of the ascending aorta and left ventricle is selected and singled out by clicking any area of the ascending aorta.Third, this image is subtracted from the original cardiac image, which leaves only the image of the coronary artery.Similar images can be created by tracking and extracting the coronary artery itself; however, tracking or extracting requires the threshold of CT attenuation, so the images can vary depending on the threshold and the ability of the workstation being used.AGV keeps the coronary artery untouched, and thus AGV does not vary among workstations.The advantage of AGV image is that it clearly demonstrates the distribution of high-density lesions, such as coronary calcifications and stents, and coronary artery stenosis in a single image, making it easier to be understood when viewed by any third party.Because the structures are displayed based upon their brightness, calcium and metal are easily distinguished from the contrast-enhanced lumen.For the detection of coronary stenosis, AGV was equally sensitive (98%) and accurate (91%) to a combination of conventional techniques, including interactive MIP and MPR as well as preset curved MPRs (98%sensitive, 94% accurate) (Jinzaki et al., 2009).AGV is also better for displaying smallcaliber segments as described in thin-slab MIP.Furthermore, AGV divided into right and left coronary artery enables viewing the lesion with the same angle as CAG, and the detection of best angle of the lesion prior to PCI (Fig. 12).Collateral arteries are also well visualized in AGV.The whole collateral arteries are demonstrated with AGV in less images than volume rendering, since AGV is a projection image (Fig. 13).Thus, AGV would be useful for explaining the severity of disease to the patient, as a guide for the reading due to its high diagnostic accuracy, and in the discussion of the treatment strategy in conference.A disadvantage of AGV is that since it is a 2D display, it requires review of multiple rotated images to understand 3D relationships.MIP does not provide in-depth information or attenuation detail within the slice, because overlapped higher-density structures obscure the lower-density structures of interest.Static MIP images can be misleading, by incorrect placement of arteries.Another disadvantage is that MIP algorithm is prone to artifacts.One well-known pitfall is the ''string of beads'' appearance, because a normal small vessel passing obliquely through a volume is only partially represented by voxels along its length (Calhoun et al., 1999).Because MIP selects the highest density voxels along a ray projected through the volume, adequate contrast enhancement is imperative.MIP has a tendency to overestimate the stenosis, especially in case with calcification (Rubin et al., 1994).Noncalcified plaque may be overlooked unless it creates significant luminal narrowing, because of its low attenuation value.

Plaque-loaded angiographic view (PLAG)
So far, there is no post-processing technique that can display an overview image of the lumen and vessel wall of a whole heart in one image (Fig. 14).PLAG is the image in which the non-calcified plaque image is merged with the AGV (Jinzaki et al., 2008).As a result, PLAG allows the noninvasive visualization of the whole coronary lumen and the vessel wall in one image.PLAG is now available in GE workstation and some other workstations.If this is not available in your workstation, it can be generated as follows: First, the color-coded plaque image of each coronary artery is created (Fig 14a).The CT attenuation of the coronary lumen is usually larger than 250HU, while that of non-calcified plaque is typically less than 120HU.Setting the color code range at less than 120 HU can generate a "plaque image", in which only the vessel wall and the non-calcified plaques are color-coded.Second, AGV image of the entire coronary artery (Fig 14b), which is similar to that shown by a CAG is separately generated, which displays the distribution of calcified plaques as well as the lumen in more detail.Third, the color-coded "plaque image" in curved MPR is re-formatted into a MIP image to merge with AGV image.Finally, loading the MIP "plaque image" onto AGV image completes the process (Figs 14c).The critical factors in PLAG are that the merged images are both in MIP format, and taking advantage of a relatively large CT attenuation discrepancy between the lumen and non-calcified plaque allows color-coding, which helps to better visualize plaques when loaded onto AGV.PLAG will provide comprehensive information for coronary CT angiography in discussing the treatment strategy of both coronary stenosis and non-stenotic non-calcified plaques.This image also enables the overview of the tortuous vessel in the chronic total occlusion (CTO), which would be useful for the percutaneous revascularization of CTO (Fig. 15).In future, with the further improvement of temporal and spatial resolution of MDCT, and with enhanced accuracy of plaque volume quantification is enhanced, this image could also be useful in evaluating and demonstrating the therapeutic effect of anti-atherosclerotic drugs in patients with multiple soft plaques in coronary arteries (the so-called vulnerable patients) on one image.

The characteristics of each post-processing
For the effective use of the various post-processing techniques, knowledge of the characteristics of each post-processing method is mandatory.The strengths and weaknesses of the available post-processing technique are summarized in Table 1.

Image interpretation
At the beginning of interpretation, an overview of the data set should be performed to detect artifacts, gross abnormalities and the presence of non-coronary cardiac or extracardiac abnormalities.Coronary artery interpretation is then performed using postprocessing images

www.intechopen.com
Image Post-Processing and Interpretation

317
Table 1.Comparison of post-processing techniques

Examination of image quality
It is imperative that an interpreter spend the first minute making sure the dataset is adequate.Artifacts due to motion, calcification, and metallic densities; image noise; and poor contrast enhancement may degrade the quality of the study as well as simulate or obscure coronary stenoses (Achenbach et al., 2003, Ferencik et al, 2003).To evaluate the image quality, the axial CT source images, multiplanar reformatted images, and also threedimensional images were analyzed.The reader should specifically state if an artery or artery segment is not interpretable and why.

Artifact
"Artifact" can be defined as any discrepancy between the reconstructed Hounsfield values in the image and the true attenuation coefficients of the object in such a way that these discrepancies are clinically significant or relevant as judged by the radiologist (Hsieh, J., 2003).Artifacts of coronary CT angiography were classified into three categories (Table 2).
The first is cardiac motion-related artifact (Choi et al., 2004).This is caused by many factors, including a heart rate greater than The second is voluntary motion that is generally preventable with careful instruction of the patient (Choi et al., 2004).Artifacts due to breathing (Fig 17) or body motion are distinctive because they affect the bones of the anterior or lateral chest wall in to the coronary arteries; these are less likely to be correctable by additional reconstructions.Oxygen supplementation may help dyspneic patients hold their breaths for a longer period.
The third is high-attenuating artifact (Choi et al., 2004).High-attenuating artifact includes, blooming artifacts, beam hardening artifacts and streaking artifacts.Blooming artifact is a result of partial-volume effects, causing the status that small high-contrast objects appear larger than they are.These artifacts commonly make calcified plaque (Fig 18a

Image noise
Improper scan parameter (low tube output for a given body size) causes poor signal-tonoise (Fig 19).This is mainly due to obesity.Improper reconstruction during a part of the cardiac cycle with reduced tube current from EKG-guided tube modulation also results in low image quality.Poor signal-to-noise is caused by low tube output for a given body size

Low contrast enhancement
The methods of reconstruction rely on Hounsfield units (HU) to make images.Values of 250 HU and above are ideal, and current protocols generally yield mean enhancement levels in the range of 300 to 350 HU for right and left coronary arteries (Johnson et al., 2009, Husmann et al., 2009).Low contrast attenuation may be secondary to improper image acquisition timing or slow contrast injection.

Non-coronary cardiac findings
Non-coronary cardiovascular structures include the pericardium, cardiac chambers, interatrial septum, interventricular septum, atrioventricular valves, ventriculo-arterial valves, pulmonary arteries, pulmonary veins, thoracic aorta, imaged aortic branch arteries, and central systemic veins.The axial images or MPRs is useful for a general understanding of the spatial relationship of these structures and abnormality.Left ventricular and left atrial myocardial walls and chamber cavities should be examined for hypertrophy, dilation, thinning (Fig. 20a), hypodense enhancement, thrombus (Fig 20b), masses, and congenital anomalies.These structures should be reviewed within the cardiac field-of-view and any abnormalities described.

Extra-cardiac structures
Extra-cardiac structures within the field of view of routine coronary CT angiography include the mediastinum, hilum, trachea and bronchi, lung parenchyma, pleura, chest wall, esophagus, stomach, liver, spleen, and colon.Interpretation of the axial images or MPR provides a general understanding of the anatomic relationships of the heart and coronary arteries with surrounding extra-cardiac structures.Review of all visible non-cardiovascular structures is important for two principal reasons: (1) recognition of primary and secondary comorbid pathology and (2) identification of findings that lead to alternative noncardiovascular diagnoses.

Coronary artery interpretation
In analyzing coronary arteries, a systematic approach using axial images and post-processing images should be followed.Realizing from the outset that no single method of analysis is without limitations, the incorporative review of the 2D images and the different 3D renderings is necessary.However, how to use the various available post-processing techniques remains a matter of personal experience and may differ among various operators.

Interpretation of axial images
The axial image evaluation is a fundamental and essential step, because information from the original data set (axial images) can be lost as a consequence of processing the images for advanced postprocessing methods.Scrolling through the axial images in the cranio-caudal direction allows interpretation of the anatomy and gross abnormality of coronary arteries.Myocardial bridge of coronary artery is often easily noticed in axial images than postprocessing images (Fig 21a).However, following each segment on each level (over 50-80contiguous images) may be difficult, and the evaluation of coronary stenosis is often more difficult compared with the post-processing methods (Jinzaki et al., 2009) (Fig 21b).For example, correct diagnosis of coronary stenosis is difficult with the axial images in the lesions located in the segment running horizontal to the axial section.Diffuse concentric plaque with positive remodeling also has a tendency to be incorrectly viewed as stenotic on axial images, possibly because the reference diameter (vessel diameter in non-diseased artery immediately proximal to the lesion) of these lesions is not measurable on the axial image.The post-processing image is better suited for evaluation of these types of lesions.

The use of post-processing images
After the interpretation of axial images, the readers should shift to post-processing images.
VR provide an overview of the coronary tree.Also, more complex anatomy (such as congenital heart disease or bypass grafts) is easily seen.For coronary artery evaluation, the combination of several types of post-processing algorithms such as thin-slab MIP, MPR and curved MPR is usually used.However, these methods require numerous images to evaluate the whole coronary artery, while AGV would be easy to overview the distribution of coronary lesion.Thus, as an initial guide for the interpretation, we recommend first to use AGV.Furthermore, in the comparison of the AGV image and combination of several types of post-processing, sensitivity was the same (98%) and there was no significant difference in accuracy (Jinzaki et al., 2009).If coronary artery stenosis is detected on AGV, the lesion should be evaluated further for vessel wall morphology and composition using curved MPR.Further evaluation could be omitted in branch vessels with no stenotic lesions on AGV, since the negative predictive value of AGV is very high (99%) and evaluating multiple diagonal branches or obtuse marginal branches is time-consuming with curved MPR or other methods.However, even if stenotic lesion is not detected in three main arteries on AGV, we recommend reconstructing curved MPR of these arteries to evaluate non-stenotic non-calcified plaque, which may be missed on AGV.This is because acute myocardial infarction is frequently related to a segment that was not significantly stenosed at previous angiography (Littele, et al., 1988, Giroud et al., 1992).Acute coronary occlusions leading to myocardial infarction tend to cluster in predictable "hot spots" within the proximal third of the coronary arteries (Wang et al., 2004).Identification of low attenuation plaques in these high-risk zones for acute coronary occlusions will be important for potentially locally directed preventive strategies.Step 1. Overview of coronary artery on angiographic view Step 2. Quantitative evaluation of coronary lesions detected on angiographic view Step 3. Evaluation of three main arteries using curved MPR and no further evaluation in branch vessels without stenotic lesions on angiographic view

The guiding principles of interpretation
To analyze coronary arteries a systemic approach should be followed.It is recommended that the coronary segments are evaluated in a sequential order from segment 1 to segment 15 following the standard report of the American Heart Association.The interpretation of coronary arteries include (1) assessment of stenosis severity, (2) vessel wall morphology and composition, (3) total occlusions, and (4) bypass graft and stent.

a. Assessment of stenosis severity
The description of luminal stenosis grading and plaque length is important in the interpretation.The degree of coronary stenosis is manually calculated by subtracting vessel diameter at the site of maximal stenosis from the mean of the vessel reference segment diameters (proximal and distal references) and then dividing the difference by the mean of the vessel reference segment diameters on curved MPR and cross-sectional image (Fig 22).
Recent cardiac-vessel analysis packages enables calculating percent stenosis automatically when the user selects the area of maximum narrowing and normal vessel proximal and distal to the stenosis (Busch et al., 2007).However, current spatial resolution of CT has not demonstrated sufficient reproducibility or accuracy in predicting findings of CAG to make such measurements a routine requirement.Thus, at the present time, Society of Cardiovascular Computed Tomography guideline recommends that arterial segments be described within broad stenosis ranges (Cheng et al., 2008, Raff et al., 2009) (Table 3), although future technical developments may improve the precision of stenosis quantification.(Raff et al., 2009).Recent studies revealed that the CT characteristics of plaques associated with ACS include positive vascular remodeling, low plaque density<30HU, and spotty calcification (Motoyama et al., 2009) (Fig. 23).Thus, the description of these features will assist interpretation.The vascular remodeling index is defined as lesion diameter/reference diameter, and is reported as positive when the diameter at the plaque site was at least 10% larger than the reference segment.Spotty calcification was defined when the calcification is <3 mm in size.Features of plaque morphology such as ulceration, dissection, and fissuring are also noted when image quality is sufficient.Coronary CT angiography may provide useful information for selecting patients suitable for PCI of CTO (Soon et al., 2007).It can show the collateral supply in the distal vessel beyond the occlusion (Fig. 13).The proximal and distal ends of an occluded segment can be distinctively seen on coronary CT angiography.Therefore, the occlusion length can be measured on CCTA (Fig. 24).Mollet et al. also reported that CTO lesion length > 15 mm measured on CT was associated with higher failure rate of PCI (Mollet et al., 2005).Heavy calcification of CTO has been reported to be an independent predictor of failed PCI of CTO (Noguchi et al., 2000).The exact location of calcified plaque on the image of coronary CT angiography may help the interventionist to steer the guide wire away from the calcified lesion that is not well seen on conventional angiography.PLAG images are able to demonstrate the overview of occluded lesion, which are helpful in providing the roadmap for interventionists during the process of wiring the occluded vessel (Fig. 15).

d. Bypass grafts, stents
The evaluation of coronary bypass grafts by coronary CT angiography is highly accurate in predicting the findings on invasive coronary angiography (Meyer et al., 2007).The location and anastomoses of bypass grafts should be described in addition to the location and severity of stenoses.The evaluation of stents by MDCT is often difficult due to blooming artifact and beam hardening effects.The evaluation of in-stent stenosis depends highly on stent size and composition.Stents 3.0 mm or larger (85%) were judged assessable; 26% of stents smaller than 3 mm were judged assessable (Schroeder et al., 2008).Different stent compositions and structures appeared to be associated with variations in the extent of residual artifact.The use of special reconstruction algorithms improved visualization of the stent lumen and reduced blooming artifact with a drawback a modest increase in noise in the images (Maintz et al., 2006).

Conclusion
Many post-processing images are available for the evaluation of coronary artery.The interpreter should know the advantage and disadvantage of each post-processing image, and perform the incorporative review of the 2D images and the different 3D renderings.

Fig. 1 .Fig. 2 .
Fig. 1.Multiplanar reformation of cardiac structure Oblique coronal image orientate better the planes of the geometrical orientations of cardiac structure (left chamber and aorta).
Fig. 3. Thin-slab maximum intensity projection of coronary stenosis When a narrowing (arrow) is suspected on axial views (a), the use of thin-slab maximum intensity projection (b) increases the overview of each coronary artery.Thin-slab maximum intensity projection enables creation of a vascular map (c) quickly without modification of several parameters.

Fig. 4 .
Fig. 4. Curved multiplanar reformation of the coronary arteries.The vessel can be seen in a long length with this technique.Fig. 4a demonstrates a multiplanar reformation of the entire right coronary artery, including a segment of posterolateral marginal branch.The atrio ventricular node artery (arrow) would require a second reconstruction to follow that artery.The left anterior descending (b), with multiple diagonal branches, often needs multiple reconstructions to demonstrate its entire course.

Fig. 5 .
Fig. 5. Curved multiplanar reformation for coronary artery stenosis Coronary stenosis (arrow) is demonstrated in the proximal segment of the right coronary artery.The observer can always create a plane orthogonal to the direction of the vessel, which allows the evaluation of the configuration of plaque and grading of a stenosis.The three orthogonal planes are at the level of a proximal normal lumen (b), a significant stenosis (c), and a distal normal lumen (d), respectively.

Fig. 6 .Fig. 8 .Fig. 9 .
Fig. 6.Volume rendering of the coronary arteries.Volume rendering provides an excellent overview of coronary anomalies, the course of coronary bypass grafts, and congenital heart disease a. Coronary anomalies: both coronary arteries originate from left coronary cusp (stent is placed in the proximal segment of left descending artery).b.Bypass grafts: right internal thoracic artery (RITA) graft communicate with the left descending artery and left internal thoracic artery (LITA) graft communicate with obtuse marginal branch.

Fig. 12 .Fig. 13 .
Fig. 12.The detection of best angle of the lesion prior to PCI The coronary stenosis (arrow) is best visualized in the view of RAO 30 and CAU 30 (middle).This information will help cardiologists in discussing treatment strategy and reduce the number of CAG image acquisitions.
Fig. 14.Plaque-loaded Angiographic View of multiple non-calcified plaques Curved MPR of the right coronary artery (a) demonstrates the non-stenotic non-calcified plaques (arrows), which are assigned blue in color-coded image (b).The angiographic view image (c) demonstrates the lumen of the entire coronary artery.The color-coded image in curved MPR is re-formatted into a MIP image and merged with the Angiographic View image.The Plaque-loaded Angiographic View image (d) demonstrates the lumen and multiple plaques (arrows) of coronary arteries in one image.
Fig. 16.Stepladder artifact due to changing heart rate during CT acquisition.Volume-rendered image (a) shows a severe motion-related artifact (arrows) caused by acceleration in heart rate toward the end of the breath hold.Selection of an optimal reconstruction window reduced the artifact (b).
Fig. 17.Stepladder artifact due to breathing Sagittal maximum intensity projection image (a) and coronal MPR image (b) show a severe respiration-related artifact in the anterior chest wall (arrow) and at the diaphragm (arrow).

Fig. 19 .
Fig. 19.Low image quality due to improper scan parameters.
Fig. 20.Non-coronary cardiac findings a. Focal thinning of myocardial wall due to ventricular diverticula (arrow) b.Thrombus in ventricular chamber (arrow)

Fig. 21 .
Fig. 21.Interpretation of axial images Myocardial bridge of coronary artery (a: arrow) is often easily noticed in axial images.The diagnosis of coronary stenosis is easy in the lesions located in the segment running the cranio-caudal direction (b: arrow), however difficult in the lesions located in the segment running horizontal to the axial section.

Fig. 22 .
Fig. 22. Workflow of post-processing and interpretation for coronary artery evaluation
Fig. 25.CT evaluation of chronic total occlusionThe total occlusion is seen in the distal segment of right coronary artery on thin-slab MIP images (a: arrows).The occlusion length can be measured on stretched MPR (b).Calcified plaque is also well visualized.

Table 2 .
Various artifacts in coronary CT angiography

Table 3 . Quantitative Stenosis Grading b. Vessel wall morphology and composition
Vessel wall morphology and composition are important information of coronary CT angiography, since these are not available on CAG.coronary CT angiography can visualize intramural presence of positively remodeled plaque and differentiate calcific, non-calcific, and mixed plaque.Society of Cardiovascular Computed Tomography recommends that the description of plaques as ''non-calcific'' is preferable to ''soft'' or ''lipid-rich'' since low CT density (in Hounsfield units) levels do not necessarily correlate closely with anatomic pathology or biochemistry