| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Innovation |
1 Department of Radiology, Kumamoto University School of Medicine, 1-1-1, Honjo,
Kumamoto, Kumamoto 860-8556, Japan. Address correspondence to T.
Nakaura.
2 Department of Cardiovascular Medicine, Kumamoto University School of Medicine,
Kumamoto 860-8556 Japan.
Received September 2, 2004;
accepted after revision January 24, 2005.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CONCLUSION. Four patients with ischemic heart disease underwent 3D volume-rendering fused images using a conversion program and volume-rendering fusion function of a computer workstation. The fusion images clearly showed the relationship of relevant coronary arteries and the abnormal perfusion territory in all patients and were useful for the evaluation of coronary artery disease.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Stress SPECT images were acquired 5 min after the injection of 201Tl (111 MBq). Coronary CT angiograms were obtained immediately after the stress SPECT study without changing the patient's position on the table. Delayed SPECT images were acquired 3-4 hr later. SPECT images were obtained using low-energy general purpose parallel-hole collimators with an ECG gating at 8 frames per R-R interval. A 35% energy window setting was used to acquire the 74- and 169-keV peaks of 201Tl, and 32 step-and-shoot images (64 x 64 matrix) were acquired over 180° from a 45° right anterior oblique to a 45° left posterior oblique angle with 1 min per projection. Transaxial images of summed raw data were then reconstructed with a maximum likelihood expectation maximization algorithm using a Butterworth filter (critical frequency, 0.6 cycle/cm, order 5). The slice thickness of each transaxial image was approximately 6 mm. Neither attenuation nor scatter correction was performed.
The CT scans were obtained using a standard ECG-gated coronary CT angiography protocol, 0.5-sec rotation time, 1.25-mm detector row width, 1.25- mm image thickness, 1.5-2.0 pitch, 120 kV, and 350 mA. A 20-mL test bolus of Omnipaque (iohexol, Daiichi Pharmaceutical Company) and a 20-mL chaser bolus of saline were used. The circulation time was determined by measuring CT attenuation values in the ascending aorta. A 100-mL bolus of contrast material was injected at 4 mL/sec; this was followed by a 30-mL chaser bolus of saline. A conventional single-sector algorithm was used when the heart rate was slower than 70 beats per minute (bpm), and a segmented reconstruction algorithm was used when it was faster than 71 bpm. For image reconstruction, we used the raw data file at 75% of the cardiac cycle. The reconstructed slice thickness was 1.25 mm; the image increment was 1.25 mm.
After image reconstruction, all CT and stress SPECT images were transferred to a PC for registration and conversion using computer software developed by one of the authors. Image registration of SPECT and CT was performed by pixel-shift manual registration using the left ventricular myocardium as an internal marker. The image corresponding to each CT slice was reconstructed from the SPECT image by linear interpolation; the output was reconstructed by the DICOM format as another series of CT studies. Original CT and reconstructed SPECT images were transferred to the workstation (ZIO M900, Zio Software) that features a volume-rendering fusion function for postprocessing. For shrinking by several pixels, the dispersion lines on registered SPECT images were trimmed.
|
|
Segmentation of the heart was performed manually from the MDCT slices. The color scale of CT angiography was that recommended for this workstation. Volume-rendering fused images were obtained from original CT and registered SPECT images using a volume-rendering fusion function. Volume-rendering fused images were compared with CT angiograms, SPECT images alone, and coronary angiograms.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
As an illustrative case, we present imaging findings of a 62-year-old man with three-vessel coronary disease. He had undergone coronary artery bypass grafting: left internal thoracic artery to left anterior descending coronary artery, left internal thoracic artery to left circumflex coronary artery, and right internal thoracic artery to right coronary artery. Three weeks later he underwent postoperative examination consisting of CT angiography and myocardial SPECT. The patency of the bypass grafts was well defined on the coronary CT angiograms. Myocardial infarction in the left anterior descending coronary artery territory was defined as a defect in tracer uptake with reverse redistribution on SPECT images (Fig. 1A). The 3D fused image showed a perfect match between the left anterior descending coronary artery territory and the tracer uptake defect (Fig. 1B).
Ischemia was suspected in another patient with a clinical history of old myocardial infarction. CT angiography and myocardial SPECT were performed preoperatively. Myocardial SPECT showed a defect in tracer uptake in the left anterior descending coronary artery territory with partial redistribution (Fig. 2A). CT angiography disclosed occlusion of segment VII and stenosis of segment IX. The 3D fused image showed good agreement between the territory of the diseased coronary artery and the myocardial perfusion defect (Figs. 2B and 2C). Conventional coronary angiography was performed for percutaneous transluminal coronary angioplasty (Fig. 2D).
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Previous studies [1-3] showed that fusion images of coronary arteries and myocardial perfusion facilitate assessment of the hemodynamic significance of coronary stenosis. Schindler et al. [1, 2] presented a method for fusing the 3D reconstructed coronary tree and 3D myocardial perfusion distribution. In these two studies, the authors subjectively determined that 74% of 162 coronary lesions from 78 patients coincided with areas of regional hypoperfusion. Faber et al. [3] developed a unification algorithm by nonlinear warping for automatically registering 3D models of the epicardial surface from perfusion SPECT and 3D coronary artery trees from coronary angiographs. The authors evaluated performance in a unique patient population in which anatomic and physiologic at-risk areas could be expected to overlap. They found an 80% and 84% overlap in at-risk and normal areas, respectively.
We propose that coronary CT angiography is more suitable than coronary angiography for image fusion. The great advantage of CT angiography is its low degree of invasiveness. Because all data are represented in true 3D formats and because the cardiac muscle is also visualized, the muscle images can be used as an internal landmark for the image. Although the resolution of coronary arteries on CT images is much lower than on standard coronary angiograms, MDCT provides clear visualization of the coronary arteries.
Although in some of our cases heartbeat and respiration led to misregistration between CT and SPECT images, the clinical evaluation was not negatively affected. To eliminate misregistration attributable to the heartbeat, we extracted SPECT data from a cardiac phase similar to that of the CT image (diastolic phase) from the ECG-gated SPECT data and used it as the myocardial perfusion data. However, the number of pixels corresponding to the thickness of the left ventricular myocardium on the extracted SPECT images was lower than that of the summed raw SPECT images. For the display of 3D volume-rendering fused images, some pixels must be eliminated from the outer side of the left ventricular myocardium. In our study it was difficult to fuse volume-rendering CT images with diastolic SPECT images because the count density in the left ventricular myocardium was seriously distorted by the elimination of the outer pixels from the left ventricular myocardium.
Although we used the same table for both coronary CT angiography and SPECT, misregistration due to respiration was observed. To resolve this problem, rotation of CT and SPECT images in any direction and registration along the cardiac axis may be necessary. This is also important in cases where registration of CT and SPECT images occurs in different rooms.
In conclusion, fused images of coronary CT angiography and myocardial SPECT are useful for the evaluation of coronary artery disease. Because these images present a panoramic view of the coronary vessels and myocardial perfusion, the relationship between the diseased artery and the myocardium can be determined without requiring difficult mental integration by the reviewer.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  O. Gaemperli, T. Schepis, I. Valenta, L. Husmann, H. Scheffel, V. Duerst, F. R. Eberli, T. F. Luscher, H. Alkadhi, and P. A. Kaufmann Cardiac Image Fusion from Stand-Alone SPECT and CT: Clinical Experience J. Nucl. Med., May 1, 2007; 48(5): 696 - 703. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
