Download pdf

Optimal reconstruction phase of ECG-gated CT angiography
in the diagnosis of acute thoracic aortic pathology (work-inprogress)
Poster No.:
R-0134
Congress:
2014 CSM
Type:
Scientific Exhibit
Authors:
D. Hocking, A. Gupta; FREMANTLE/AU
Keywords:
Cardiovascular system, Emergency, CT, CT-Angiography,
Technical aspects, Artifacts
DOI:
10.1594/ranzcr2014/R-0134
Any information contained in this pdf file is automatically generated from digital material
submitted to EPOS by third parties in the form of scientific presentations. References
to any names, marks, products, or services of third parties or hypertext links to thirdparty sites or information are provided solely as a convenience to you and do not
in any way constitute or imply RANZCR/AIR/ACPSEM's endorsement, sponsorship
or recommendation of the third party, information, product or service. RANZCR/AIR/
ACPSEM is not responsible for the content of these pages and does not make any
representations regarding the content or accuracy of material in this file.
As per copyright regulations, any unauthorised use of the material or parts thereof as
well as commercial reproduction or multiple distribution by any traditional or electronically
based reproduction/publication method ist strictly prohibited.
You agree to defend, indemnify, and hold RANZCR/AIR/ACPSEM harmless from and
against any and all claims, damages, costs, and expenses, including attorneys' fees,
arising from or related to your use of these pages.
Please note: Links to movies, .ppt slideshows, .doc documents and any other multimedia
files are not available in the pdf version of presentations.
Page 1 of 13
Aim
Helical multi-detector row computed tomography (MDCT) has an established role in
the diagnosis of acute aortic pathology. Motion-free visualisation of the aortic root
and ascending aorta allows the radiologists to make a confident diagnosis and can
aide with surgical planning. Retrospectively ECG-gated scans acquire complete data
throughout the cardiac cycle and are reconstructed at a specified point in the R-R interval,
significantly reducing motion artifact [1]. Optimal reconstruction phase varies depending
on the structure being evaluated and the patient's heart rate. Figures 1-3 illustrate the
difference in quality between ungated CT (Fig. 1 on page 2), incorrect phase of gated
CT (Fig. 2 on page 3) and correct phase of gated CT (Fig. 3 on page 4) at similar
levels in the same patient.
Current protocol at our institution is to reconstruct a 75% phase when the heart rate is
less than 80 beats per minute (bpm) and a 40% phase for heart rates greater than 80
bpm; a practice which is based on operator experience. Similar past research by MorganHughes et al. has suggested a statification threshold at 70 bpm, with best phases at
75% and 50%, respectively [2]. However, this prior study utilised older equipment and the
50% phase did not reach statistical significance. This study aims to test the hypothesis
that images are optimally reconstructed either during the end-systolic phase (35-45%)
or the mid-diastolic phase (70-80%) and to establish the heart rate threshold where the
transition occurs. The results will serve to form the basis of a standardized, evidencebased protocol for departmental use.
Images for this section:
Page 2 of 13
Fig. 1: Selected, non-consecutive axial slices from the level of the tubular ascending
aorta (top left) to the aortic root (bottom right) in an un-gated CT pulmonary angiogram
performed in an individual patient (Patient 8). The heart rate was not recorded at the time
of the scan. Images demonstrate extensive motion artifact at all levels, which could be
misinterpreted as an intimal flap (arrows).
Page 3 of 13
Fig. 2: Selected, non-consecutive axial slices from the level of the tubular ascending
aorta (top left) to the aortic root (bottom right) in a retrospectively gated CT angiogram
targeted to the thoracic aorta. This was performed in the same individual patient shown
in Figure 1 (Patient 8), shortly after the ungated CT scan. The heart rate was 75 beats per
minute at the time of imaging. Images have been reconstructed at the 75% R-R interval.
The presence of motion artifact is reduced compared to the ungated study, but remains
present to a lesser degree (arrows).
Page 4 of 13
Fig. 3: Selected, non-consecutive axial slices from the level of the tubular ascending
aorta (top left) to the aortic root (bottom right) in a retrospectively gated CT angiogram
targeted to the thoracic aorta. This was performed in the same individual patient shown
in Figure 1 (Patient 8), shortly after the ungated CT scan. The heart rate was 75 beats per
minute at the time of imaging. Images have been reconstructed at the 40% R-R interval.
Almost no motion artifact is visible.
Page 5 of 13
Methods and materials
Equipment and CT Scan Parameters
All patients were scanned on a 64-slice multi-detector row CT scanner (Philips Brilliance
64 running under the V.3.5.5.5 2007 11 Aprl 2013 software ugrade). The scan protocol
included dual scout views, followed by un-enhanced thoracic CT and subsequent
contrast enhanced aortic angiogram from the aortic arch to the aortic bifurcation, using a
retrospective ECG-gating technique. Contrast phase scan parameters were as follows:
64 slice helical mode, collimation 0.625mm, 512 pixel matrix, pitch 0.299mm, gantry
rotation time 0.4 seconds, FOV 350 and 120 kV tube voltage. Tube current varied by
patient size, ranging between 300-600 mAs. Iodinated intravenous contrast was injected
with an automated power injector. Optimal contrast timing was achieved with bolus
tracking. For the purposes of this study, axial images were reconstructed at 3mm slice
thickness at 10% intervals between 20-80% of the R-R interval. Additional phases at
35%, 45% and 75% were also included to more closely investigate end-systolic and middiastolic phases. Quality assessment was determined exclusively from the reconstructed
axial data. The complete CT study data was transfered to a departmental research harddisc drive, which was utilised exclusively during image analysis.
Patient Selection
Consecutive patients ungergoing electrocardiogram-gated (ECG-gated) CT thoracic
angiography, performed on the institutional emergency department CT scanner, during
the trial period between 15th February 2014 and April 20th 2014, were included in study
data collection. One patient was later excluded due to uniformly high quality images
throughout all phases, limiting assessment for "best phase."
Data Gathering
A unique, anonymized patient identifier was issued to each patient. Age and gender
were recorded. The mean patient heart rate during the CT scan was recorded for
each patient, as obtained from the ECG performed by the CT scanner at the time
of imaging. The heart rate regularity and presence of acute aortic pathology were
noted. The images were subjectively evaluated for the "best phase" as defined by the
phase with the least overall movement artifact at the aortic root and within the tubular
ascending aorta. As the workstation software presented the phases in order of phase,
the investigators were not blinded to the phases at the time of scoring. Assessment was
by two independant investigators, one radiology registrar with 3 years of experience and
an experienced thoracic radiologist. Consensus on disconcordant results was achieved
through discussion. Tabulated findings were recorded in a Microsoft excel spreadsheet
stored on an institutional research drive.
Page 6 of 13
Statistical Analysis
Statistical analysis was performed by an experienced biostatistician using the Stata/
SE v13.1 software package. Inter-rater agreement and reliability were assessed using
the intraclass correlation coefficient and mixed-effects ML regression, respectively.
Correlation between heart rate and best phase was assessed via random effects logistic
regression analysis to evaluate overall relationship. Two-sample Wilcoxon rank-sum test
was performed to establish significance between high and low heart-rate groups.
Results
A total of 20 patients were scanned during the study period. The majority of patients
were male (15/20; 75%). The mean age was 58.0 years (SD 12.2, range 25-76). The
mean heart rate across the study population was 63.5 beats per minute (SD 17.6, range
46-101). All patients included in the study had a regular heart rate. Acute aortic pathology
was present in 3 cases. Patient demographics and observer findings are provided in
Table 1 on page 8. Summary of results are presented in Table 2 on page 8.
Inter-Rater Reliability
The inter-rater reliability was good, with an intraclass correlation coefficient of 0.63 (95%
CI 0.34 - 0.84). There was no difference in the reliability of ratings based on mixed-effects
ML regression (p = 0.903). In one case (Patient 14), one investigator did not generate
an opinion as to the best phase due the impression that all phases in the study were of
equally high quality. This patient was excluded from "best phase" analysis as consensus
could not be reached. Of the remaining cases, difference of opinion occured in 13 of 19
cases (68%). Of these disagreements, 3/13 (8%) were due to major variance of opinion
between end-systolic and mid-diastolic "best phase". At consensus, in all three cases the
variance was due to motion being minimal at the aortic root during the end-systolic phase,
but minimal in the tubular ascending aorta during the mid-diastolic phase (Fig. 4 on page
9), with differening initial opinion as to which constituted better overall image quality.
At consensus, the overall best image quality was determined to be during mid-diastole. In
5/13 cases (38%), disagreement was due to a lack of perceptible difference between the
phases initially selected by the investigators. In these cases, consensus reflected both
phases, as neither could be considered of superior quality. In the remaining 5/13 (38%),
minor differences were visible, but did not alter the preferred phase group.
Best Phase of Reconstruction
Analysis was performed on mixed investigator scores (n = 38) and consensus results (n
= 19). Pictorial representations of individual and consensus results are provided in Fig. 5
Page 7 of 13
on page 10 and Fig. 6 on page 11, respectively. There was a significant correlation
between heart rate and best phase (p = 0.015; random effects logistic regression). The
best phase fell into two significantly different groups (p < 0.0001; Wilcoxon signed-rank
test), summarised in Table 2 on page 8. At higher heart rates (mean 88.6 bpm, range
75 - 101, SD 13.0), the mean "best phase" was 39% (range 30-45%) correlating with endsystole. At lower heart rates (mean 56.3 bpm, range 46 - 76, SD 8.8), the mean "best
phase" was 74% (range 70-80%) correlating with mid-diastole. Based on the available
data, a best estimate threshold value between the two groups is at 75 bpm. However,
the sample size is too small to confirm this value.
Images for this section:
Table 1: Tabulated data from all patients including anonymised patient identifier, age,
gender, individual investigator "best phase" preference and consensus results. * indicates
an absent score. A preference was not indicated by investigator 2 in this patient due to
uniformly high image quality across all phases. A consensus was not reached.
Page 8 of 13
Table 2: Summary of final results derived from the study data. Values in parentheses
indicate percentages.
Page 9 of 13
Fig. 4: Selected, non-consecutive images obtained during retrospectively ECG-gated CT
angiogram of the thoracic aorta (Patient 19). Images on the left demonstrate the 80%
phase and images on the right demonstrate the 40% phase. Motion artifact is present in
the 80% phase at the level of the aortic root (bottom images), but not seen in the tubular
aorta (top images). The contrary is true in the 40% phase (right images). Arrows indicate
location of wall motion artifact.
Page 10 of 13
Fig. 5: Scatterplot showing mixed investigator scores for best phase of reconstruction (yaxis) relative to heart rate (x-axis). Investigator 1 scores are shown in blue. Investigator
2 scores are shown in green. The distribution suggests a bimodal relationship.
Fig. 6: Scatterplot showing consensus for best phase of reconstruction (y-axis) relative
to heart rate (x-axis). The distribution suggests a bimodal relationship with a transition
heart rate threshold in the region of 75 beats per minute.
Page 11 of 13
Conclusion
This study demonstrates a significant correlation between heart rate and optimal phase
of image reconstruction at retrospectively ECG-gated CT angiography of the thoracic
aorta (p = 0.015). The best phase appears to correlate with two statistically significant
groups at end-systole and mid-diastole (p < 0.0001). Based on these data, reconstruction
recommendations are as summarised in Table 3 on page 12. Where the patient heart
rate is less than 75 beats per minute, the data should be reconstructed during the middiastolic phase (near to 75%). When the heart rate is above this threshold, the data should
be reconstructed in the end-systolic phase (near to 40%). Due to the small sample size,
further study is required to confirm the 75 bpm threshold.
Images for this section:
Table 3: Summary of recommendations derived from the study results. The 75 beats per
minute threshold is provided as a best estimate, based on the available data.
Page 12 of 13
Personal information
Dr David Hocking is a third year trainee registrar in the Western Australia Radiology
Training (WART) program, based out of Royal Perth Hospital.
Dr Ashu Gupta is a consultant radiologist at Fremantle Hospital in Perth, Western
Australia. He has a special interest in chest and cardiac radiology.
References
1.
2.
Roos JE, Willmann JK, Weishaupt D, Lachat M, Marincek B, Hilfiker PR.
Thoracic aorta: motion artifact reduction with retrospective and prospective
electrocardiography-assisted multi-detector row CT. Radiology. 2002
Jan;222(1):271-7.
Morgan-Hughes GJ, Owens PE, Marshall AJ, Roobottom CA. Thoracic
Aorta at Multi-Detector Row CT: Motion Artifact with Various Reconstruction
Windows. Radiology. 2003 Aug;228(2):583-8.
Page 13 of 13