| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
Department of Radiology, Ehime University School of Medicine, Shitsukawa,
Shigenobu-cho, Onsen-gun, Ehime 791-0295 Japan.
2
Department of Radiology, Ehime-Imabari Hospital, 4-5-5, Ishii-cho,
Imabari-city, Ehime 794-0006 Japan.
3
Department of Cardiology, Ehime-Imabari Hospital, Imabari-city, Ehime 794-0006
Japan.
Received December 21, 1998;
accepted after revision June 23, 1999.
Address correspondence to T. Mochizuki.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. With a single breath-hold, the patient's entire heart was scanned with an ECG-gating technique (3-mm-thick collimation, 2 mm per rotation table speed, 0.8 sec per rotation, and 50 rotations through 10 cm in total). Using a 0.2-mm (0.08-sec) interval (10 slices per rotation) overlapping reconstruction, about 500 axial slices were obtained and reordered to separate different cardiac cycles. Then, 2D cardiac axes and 3D images were reconstructed and animated movies of the 2D and 3D images were produced. In 21 patients, the left ventricular end-diastolic volume, end-systolic volume, and ejection fraction were assessed and compared with left ventriculography. Correlations and agreements between CT and left ventriculography were determined.
RESULTS. Close correlations between CT and left ventriculography were obtained (r = 0.95, 0.98, and 0.91, for end-diastolic volume, end-systolic volume, and left ventricular ejection fraction, respectively; p << 0.0001 for all values). The limits of agreement between CT and left ventriculography were 44.3 to -44.5 ml for end-diastolic volume, 19.8 to -29.0 ml for end-systolic volume, and 19.7% to -9.5% for left ventricular ejection fraction.
CONCLUSION. This cardiac application of helical CT provides a clear morphology along the cardiac axes and 3D images and an assessment of left ventricular volumes (end-diastolic volume, end-systolic volume, and left ventricular ejection fraction).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Using CT ventriculography of two-dimensional (2D) cardiac axes and 3D images, we have described a case of myocardial infarction with thin wall and left ventricular thrombus [9]. We also have described a case of acute myocardial infarction detected as deficient myocardial enhancement in the early phase and as enhancement in the delayed phase [10]. We sought to determine if CT ventriculography is accurate relative to conventional ventriculography for measuring end-diastolic volume, end-systolic volume, and left ventricular ejection fraction. To reproduce CT ventriculography and to validate its usefulness for future studies, a whole picture of CT ventriculography is proposed and the potential merits are discussed.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
To evaluate the consistency of enhancement throughout the left ventricle, regions of interest of 1 cm in diameter were drawn at the upper fourth, mid ventricular, and the lower fourth levels of the left ventricular cavity, using end-diastolic phase axial images. As a reference, a region of interest of 0.5 cm in diameter was also drawn in the septum at the same three levels. Then, mean attenuation numbers (H) in the region of interest at the three levels were compared.
The angiographic system used for conventional left ventriculography was the biplane Integris V3000 (Phillips Medical Systems, Best, the Netherlands) and the cardiac function analyzing system, Cardio 500 (Kontron Electronik, Munich, Germany). During the injection of the contrast medium (30 ml of 320 mg I/ml at 10 ml/sec), biplane images were obtained at a rate of 30 images per second. Although biplane filming was performed for imaging, only the right anterior oblique view was analyzed to calculate end-diastolic volume, end-systolic volume, and left ventricular ejection fraction, using the area-length method. The end-diastolic and end-systolic phases were defined visually.
To investigate how the heart rate affects the measurements of end-diastolic volume, end-systolic volume, and left ventricular ejection fraction with CT ventriculography, differences (left ventriculography values — CT values) were evaluated as a function of heart rate.
To assess the radiation dose, the CT chamber
(Physikalisch-Techiniche-Werkstaetten Dr. Pychlau, Freiburg, Germany) was
placed in the center of the MIX-DP phantom (a cylinder-shaped phantom with
small holes for the CT chamber: 
= 1.01, 
= 30 cm; Kondoh
Chemical Industry, Kobe, Japan). The phantom was scanned with the same
parameters used for CT ventriculography (3-mm-thick collimation, 2 mm per
rotation table speed through 10 cm, 140 kV for tube voltage, and 160 mA for
tube current) and with the routine parameters for whole-lung scanning
(10-mm-thick collimation and 10 mm per rotation table speed through 30 cm, 140
kV, and 160 mA).
Data Acquisition and Processing
The patients laid supine on the CT table, raised their hands over their
head, and wore rubber gloves for insulation. ECG leads were placed on both
wrists and the left forearm to avoid artifacts. After a scout view for
positioning, the contrast-enhanced CT data were acquired during a single
breath-hold. Patients inhaled 3 l/min of oxygen before the acquisition while
the scan parameters were established. Fifty seconds after IV injection of the
contrast medium (100 ml of 320 mg I/ml at 1.2 ml/sec) began, the scan was
started with the ECG-gating technique. The exact start and end points of the
scan were indicated on the ECG record. The parameters used were 0.8 sec per
rotation, 3-mm-thick collimation, and 2 mm per rotation table speed (about 50
rotations through 10 cm during 40 sec). Voltage and electric current were 140
kV and 160 mA, respectively. The raw data were reconstructed at 0.2-mm
(0.08-sec) intervals (about 500 axial slices). Because a half-scan helical
reconstruction algorithm was applied, the time resolution (full width at half
maximum [FWHM]) to obtain one axial slice was 0.4 sec. The 180° helical
reconstruction algorithm used in this study and the conventional 360°
reconstruction algorithm are illustrated to clarify the difference of the time
resolution (Fig. 1).
|
All axial slices (about 500) were transferred to the workstation at which further processes were performed. Data sets of different cardiac cycles using 10 slices per rotation of overlapping reconstruction were extracted (Fig. 1). When the center of the reconstruction for the slice was on the R wave or nearest to the R wave, the slice was defined as the end-diastolic phase. Then the slices on R waves were listed for extraction of the end-diastolic phase data set. The framing number of the R-R interval was generated as follows: mean R-R interval (msec) / 80 (msec), rounded to integers (about 10-15 phases per cardiac cycle, depending on the heart rate). For example, 10 phases were extracted in patients with a mean heart rate of 75 beats per minute (R-R = 800 msec) during the acquisition and 15 phases were extracted in patients with a mean heart rate of 50 beats per minute (R-R = 1200 msec). We dealt with beat-to-beat variability by substituting the adjacent phase slice for the deficient phase slice when the phase number was smaller than the average and by discarding the extended phase slice when the phase number was larger than the average (Fig. 1). The next phase of the end diastole was obtained by summing all [end diastole + 1] frames. Thus, data sets of all phases were extracted. The end-systolic phase was detected visually by paging through mid ventricular axial slices. In each phase, 2D images in cardiac axes (vertical long, horizontal long, and short axes) were reformatted to evaluate the myocardium and the ventricles.
Heart cavities were extracted from the data by applying a specific threshold to render 3D images. The threshold to delineate the endocardial border was visually defined by referring to the mid ventricular axial slice, because the appropriate threshold depends on the magnitude of enhancement (threshold attenuation varied from 110 to 160 H). Axial slices of the end-diastolic and end-systolic phases were used to assess end-diastolic and end-systolic volumes. The left ventricular boundaries of the end-diastolic and end-systolic slices were delineated manually, using the standard packaged software. Then, the end-diastolic and end-systolic volumes were assessed by summing all the left ventricular cavities using the axial slices of end-diastolic and end-systolic phases. The left ventricular ejection fraction was calculated from the end-diastolic and end-systolic volumes. Correlation and agreement of left ventricular values (end-diastolic volume, end-systolic volume, and left ventricular ejection fraction) between CT ventriculography and left ventriculography were investigated.
Statistical Analysis
Statistical differences among the mean CT numbers (H) of the upper fourth,
mid ventricular, and the lower fourth levels were evaluated using analysis of
variance. A probability value of less than 0.05 was considered significant.
Left ventricular values (end-diastolic volume [ml], end-systolic volume [ml],
and left ventricular ejection fraction [%]) were expressed as mean ±
SD. Correlation coefficients (r) were computed by comparing the
values from CT ventriculography and left ventriculography; if r was
significant (p << 0.05), the limits of agreement were calculated
according to the method of Bland and Altman
[11]. Correlation coefficients
between the differences (left ventriculography — CT values) and the
heart rate were also computed.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CT attenuation at the upper fourth, mid ventricular, and lower fourth levels was 199.2 ± 33.2 H, 198.9 ± 38.3 H, and 200.5 ± 33.3 H in the left ventricular cavity, and 101.1 ± 14.5 H, 99.9 ± 12.8 H, and 102.1 ± 15.7 H in the myocardium, respectively (mean ± SD, not significant among the three levels) (Fig. 2). It took approximately 90 min to complete the reconstruction in all phases (approximately 30 min for extraction of all phases and approximately 3 min for 2D and 3D reconstruction in each phase). Observation of wall motion and systolic thickening was possible in all patients. Representative normal 2D and 3D images in end-diastolic and end-systolic phases showed clear morphologic information about the left ventricular cavity and myocardium (Fig. 3A), (Fig. 3B). Six of 21 patients could not hold their breath for 40 sec. In two of the six, reconstructed images were distorted and showed bending or steplike artifacts. End-diastolic volume, end-systolic volume, and left ventricular ejection fraction of these two patients with distorted images were within the range of regression line ± 1 SD. Because the remaining four patients restarted their breathing during the last few seconds, the whole left ventricle was covered during the breath-hold. Left ventricular values of the CT ventriculography were closely correlated with those of left ventriculography (Figs. 4 and 5). However, the limits of agreement between CT ventriculography and left ventriculography were wide (44.3 to -44.5 ml for end-diastolic volume, 19.8 to -29.0 ml for end-systolic volume, and 19.7% to -9.5% for left ventricular ejection fraction) (Table 1). The difference (left ventriculography — CT) and heart rate showed no significant correlation in the measurement of end-diastolic and end-systolic volumes (r = -0.04 and p = 0.881 for end-diastolic volume and r = -0.19 and p = 0.422 for end-systolic volume, respectively). However, as the heart rate increased, left ventricular ejection fraction of CT ventriculography became smaller than that of left ventriculography (r = 0.41, p = 0.065).
|
|
|
|
|
|
The radiation dose that simulated the CT ventriculography (50 rotations, 3-mm-thick collimation, and 2 mm per rotation through 10 cm) was 184 mGy for every 10 cm, and the radiation dose that simulated the routine whole-lung scan (30 rotations, 10-mm-thick collimation, and 10 mm per rotation through 30 cm) was 374 mGy for every 30 cm.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Method differences in assessing left ventricular volumes, such as summing
of 2D images (
area x thickness = 3D method) with CT
ventriculography versus the area—length method with left
ventriculography, may have caused disagreement. Because only the right
anterior oblique view is applied to calculate end-diastolic volume and
end-systolic volume in left ventriculography using the area—length
method, the geometric assumption is less accurate for the gold standard.
Although CT ventriculography and left ventriculography were performed within 2
weeks of each other, we cannot determine if the true end-diastolic volume,
end-systolic volume, and left ventricular ejection fraction differed between
studies. Therefore, the disagreement does not necessarily indicate that the CT
ventriculography was less accurate. The 3D volume assessment may be more
accurate than the area—length method
[12].
Many factors may affect the accuracy of left ventricular measurements using CT ventriculography, such as time resolution (FWHM), heart rate, slice thickness (collimation), framing number (phases) of the cardiac cycle, left ventricular shape and size, methods for delineating the left ventricular border, and duration of breath-hold. Pulsating phantom studies and clinical comparisons with other techniques, such as MR imaging, sonography, electron beam CT, and nuclear imaging, are necessary to determine whether CT ventriculography (or the 3D method) is more accurate than left ventriculography (or 2D method) for measuring left ventricular volumes. Comparison of CT ventriculography with 3D maximum-intensity-projection images and left ventriculography in the same projection is essential in future studies.
CT ventriculography provided 2D and 3D images of the entire heart in different cardiac cycles, such as the end-diastolic phase and the end-systolic phase with few motion artifacts (Fig. 3A), (Fig. 3B). Morphologic evaluation of the myocardium and the ventricular cavity was possible with a better orientation using 2D images in cardiac axes. The interval between views (axial slices) within a given phase of the cardiac cycle often had different relative table distances from beat to beat. Skips or gaps between views were rarely seen in patients with a slow heart rate (<<50 beats per minute). The workstation software could deal with the interval variances and could interpolate the skips, yielding negligible artifacts. Therefore, we felt that the slice interval variances would have negligible or minimal effects on the measurement of left ventricular volumes.
Animation allows assessment of wall motion and systolic thickening and quantification of wall motion, wall thickness, and systolic thickening, but additional studies are needed to evaluate accuracy in the assessment of wall motion and systolic thickening.
Contrast enhancement was consistent throughout the left ventricular cavity and the contrast dose was sufficient for the threshold method to extract left ventricular cavities in all 21 patients. Proper IV injection rate of the contrast material and scan timing are important to obtain good enhancement and to avoid artifacts from the superior vena cava. IV administration of the contrast medium (100 ml at 1.2 ml/sec) would cause minimal hemodynamic changes in cardiac output compared with the volume of contrast medium for left ventriculography (30 ml at 10 ml/sec).
Evaluation of myocardial perfusion is one of the potential merits of CT ventriculography. We previously visualized deficient myocardial enhancement of acute myocardial infarction [10]. Paging through the original axial slices was helpful in distinguishing true myocardial perfusion defects from artifacts in patients with acute myocardial infarction when a low-density area of undetermined origin was observed. Artifacts were linear, whereas perfusion defects of infarct were arced or rounded and were observed consistently throughout the infarct. Direct evaluation of myocardial wall enhancement, which has been evaluated by electron beam CT [13, 14, 15], may be used to detect coronary artery disease.
To complete the procedures manually took 90 min. Most of the time was needed to extract all phases and to load, sort, and file the data. Sophisticated software, which can handle data sequentially within a few minutes, was necessary for clinical implementation. Development of interactive 2D and 3D animation will help accelerate the clinical use of CT ventriculography.
The radiation dose of CT ventriculography for the 500 slices at 0.2-mm intervals (10 slices per rotation) is the same as that from 50 slices with 2-mm intervals (one slice per rotation). The radiation dose of CT ventriculography (18.4 mGy/cm x 10 cm) was one half of the simulated whole-lung helical CT (12.5 mGy/cm x 30 cm), whereas the average skin dose of biplane left ventriculography was approximately 90 mGy in this study. Therefore, radiation dose was not a limitation.
For patients who cannot hold their breath for 40 sec even after oxygen inhalation, faster imaging protocols (such as 3-mm-thick collimation and 3 mm per rotation table speed [27 sec] or 5-mm-thick collimation and 3-4 mm per rotation table speed [27-20 sec]) should be considered. This problem will likely be resolved by the new subsecond multidetector helical CT.
The main limitation of CT ventriculography was the time resolution for obtaining one axial slice. With the half-scan helical reconstruction algorithm, we applied one of the shortest time resolutions (0.4 sec) reported to our knowledge for helical CT. With this time resolution, the end-diastolic and end-systolic phase images were clear with fewer motion artifacts, but some blurring occurred in the rapid contracting phase. The use of a ß-blocker may help reduce the heart rate and motion artifacts. Some CT manufacturers have released new subsecond helical CT scanners (0.5-0.8 sec per rotation) with multirow detector systems. FWHM of 0.25 sec is ready for clinical use. A new reconstruction algorithm for the multirow detector CT may be developed to shorten the FWHM. The multidetector subsecond helical CT will also shorten the acquisition time by up to one sixth (6 sec to scan the entire heart) with the same parameters that we used.
In summary, the merits of CT ventriculography are that the morphology of the whole heart cavity and myocardium in cardiac axes, 3D images, and animation of these 2D and 3D images can be generated and that left ventricular values (end-diastolic volume, end-systolic volume, and left ventricular ejection fraction) can be assessed. Potential merits are that myocardial perfusion, wall motion, and systolic thickening can be evaluated, and left ventricular mass can be assessed. Further research is necessary to determine other potential merits.
The new application of helical CT we have outlined allows assessment of left ventricular values (end-diastolic volume, end-systolic volume, and left ventricular ejection fraction). CT ventriculography will likely become more practical using multidetector subsecond helical CT and sophisticated software, which can handle data within a few minutes.
Acknowledgments
We thank Yun Shen, General Electric Yokogawa Medical Systems, for technical
and physiologic advice concerning ECG-gating and helical reconstruction
algorithms.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Nagao, H. Matsuoka, H. Kawakami, H. Higashino, T. Mochizuki, K. Murase, and M. Uemura Quantification of Myocardial Perfusion by Contrast-Enhanced 64-MDCT: Characterization of Ischemic Myocardium Am. J. Roentgenol., July 1, 2008; 191(1): 19 - 25. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Minami, H. Chung, M. Kudo, S. Kitai, S. Takahashi, T. Inoue, K. Ueshima, and H. Shiozaki Radiofrequency Ablation of Hepatocellular Carcinoma: Value of Virtual CT Sonography with Magnetic Navigation Am. J. Roentgenol., June 1, 2008; 190(6): W335 - W341. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Utsunomiya, K. Awai, T. Sakamoto, T. Nishiharu, J. Urata, A. Taniguchi, T. Nakaura, and Y. Yamashita Cardiac 16-MDCT for anatomic and functional analysis: assessment of a biphasic contrast injection protocol. Am. J. Roentgenol., September 1, 2006; 187(3): 638 - 644. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Koyama, H. Matsuoka, T. Mochizuki, H. Higashino, H. Kawakami, S. Nakata, J. Aono, T. Ito, M. Naka, Y. Ohashi, et al. Assessment of Reperfused Acute Myocardial Infarction with Two-Phase Contrast-enhanced Helical CT: Prediction of Left Ventricular Function and Wall Thickness Radiology, June 1, 2005; 235(3): 804 - 811. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. P. Lawler, D. Ney, H. K. Pannu, and E. K. Fishman Four-Dimensional Imaging of the Heart Based on Near-Isotropic MDCT Data Sets Am. J. Roentgenol., March 1, 2005; 184(3): 774 - 776. [Full Text] [PDF]
|
|
|
|
|
|
M. Yamamuro, E. Tadamura, S. Kubo, H. Toyoda, T. Nishina, M. Ohba, R. Hosokawa, T. Kimura, N. Tamaki, M. Komeda, et al. Cardiac Functional Analysis with Multi-Detector Row CT and Segmental Reconstruction Algorithm: Comparison with Echocardiography, SPECT, and MR Imaging Radiology, February 1, 2005; 234(2): 381 - 390. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Mochizuki, T. Ohtani, H. Higashino, Y. Sugawara, T. Tsuda, M. Sekiya, M. Miyagawa, K. Ohmoto, and J. Ikezoe Tricuspid Atresia With Atrial Septal Defect, Ventricular Septal Defect, and Right Ventricular Hypoplasia Demonstrated by Multidetector Computed Tomography Circulation, November 14, 2000; 102 (20): e164 - e165. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
