Outline
Effect of Scan Parameters in Cardiac
Imaging with MDCT
•
•
•
•
•
•
Mahadevappa Mahesh, M.S., Ph.D.
The Russell H. Morgan Department of Radiology
and Radiological Science
Johns Hopkins University, Baltimore, MD
Introduction
Fundamentals of Cardiac CT Imaging
Temporal and Spatial Resolution
Pitch, Geometric Efficiency,…
Effect of scan parameters on image quality
Conclusions
47th Annual Meeting, Seattle, WA (Yr 2005)
Essentials for Cardiac Imaging
Key Issues in Cardiac Imaging with CT
•
•
Fast imaging - High temporal resolution to
coronary arteries located close to heart muscles
freeze cardiac motion and avoid artifacts
•
that show strong movement during cardiac cycle
•
•
•
Fine imaging - High spatial resolution to
resolve small lesions in any plane
•
Radiation dose - the consequences of CT
Diastolic Phase versus Heart Rate
400
200
Diastole
Exposure
time
Least cardiac motion is
observed during diastolic phase
•
Diastole phase narrows with
increasing heart rate
•
Desired temporal resolution for
motion free cardiac imaging
100 ms
0
40
•
500 ms
250 ms
ms
Rapid movement is present during systole phase
Imaging should be performed during diastole phase
Image acquisition and reconstruction are to be
synchronized accurately with heart movement
imaging
600
High temporal resolution is key for imaging
60
80
100
Heart rate (bpm)
120
•
−
~ 250 ms for heart rates ~ 70 bpm
−
~ 150 ms for heart rates ~ 100 bpm
Motion-free imaging during
other phases requires temporal
resolution ~50 ms
Mahadevappa Mahesh, MS, Ph.D.
Johns Hopkins University, Baltimore, MD
Essentials for Cardiac Imaging
•
Most proximal coronary segments
(RCA, LAD) require high submillimeter isotropic spatial resolution
•
Sufficient contrast-to-noise ratio is
required to resolve small and lowcontrast structures such as plaques
•
Low-contrast resolution with limited
radiation exposure at shortest
exposure time is key
RCA
LAD
1
High Quality Coronary CTA images
• Cardiac imaging is a high demanding
application of CT
• Temporal, spatial and contrast
resolution are to be optimized and
Axial
Coronal
Sagittal
also radiation exposure are to be limited
Cardiac Images from 16 section MDCT*
Temporal Resolution
*Mahesh M, Clini Cardio Vasc Img Textbook, pp 1-77, 2004
Prospective ECG Triggering
Approaches for stopping heart motion
•
Temporal resolution 200 – 250 msec
Radiation dose minimized
Limited data set
Acquire all data fast enough to stop
cardiac motion
•
Achieved in MDCT by
−
Prospective ECG triggering
−
Retrospective ECG gating
Conventional Axial “ Partial Scan ” (Step and Shoot)
Preset
Delay
X-ray ON
moving
couch-top
Preset
Delay
X-ray ON
ECG
Mahadevappa Mahesh, MS, Ph.D.
Johns Hopkins University, Baltimore, MD
2
Partial vs Segmented Reconstruction
Retrospective ECG Gating
Temporal Resolution 200 - 250 msec
Radiation dose higher than
prospective triggering
㻨㻦㻪
Continuous spiral scan
Retrospective
reconstruction
(partial scan:
180o + fan angle)
Continuous recording of spiral scan and ECG
TR: 200 msec
䝿䚭䝿䚭䝿
ECG
moving
couch-top
Time / Pos.
䝿䚭䝿䚭䝿
䝿䚭䝿䚭䝿
Time/position
Segmented
reconstruction
Factors affecting Temporal Resolution
•
•
Gantry rotation speed
Prospective triggering
−
Retrospective gated
• Partial or segmental
•
•
Pitch
•
Partial scan reconstruction
−
•
continuous segment of projection
data at single heartbeat
Segmented reconstruction
−
Post-processing algorithms
Different segments of projection
data from same phase of cardiac
cycle at successive heartbeats used
Partial Scan Reconstruction
•
•
•
Data from prescribed time range during one
cardiac cycle is selected for reconstruction
Multisegment Reconstruction
•
Each heart cycle provides segment of data
required for partial scan reconstruction
•
For ‘M’ segments (‘M’ heart cycles), maximum
temporal resolution is
200 to 270 ms temporal resolution is
achieved for 0.4 s gantry rotation
Works best with very low pitch (p<0.25)
•
Mahadevappa Mahesh, MS, Ph.D.
Johns Hopkins University, Baltimore, MD
TR: <100 msec
Retrospective Reconstruction
Image Reconstruction
−
䝿䚭䝿䚭䝿
−
‘TR/2M’ where TR is gantry rotation time
−
100 ms with 2 segments and 50 ms with 4 segments
for TR of 0.4 sec
For M segment, temporal resolution vary
between TR/2 to TR/2M
3
Temporal Resolution:
Partial vs Segmented ECG gated reconstruction
Temporal Resolution
•
Depends on gantry rotation time
-
•
0.5 - 0.37 second with 64 section MDCT scanners
Up to 80 - 250 ms achieved through partial scans
or sub-segment data reconstruction
•
Improves with sub-segment data reconstruction,
but spatial resolution decreases and motion
Half-scan reconstruction
Temporal resolution: 250 msec
Segmented reconstruction
Temporal resolution: ~105 msec
artifacts increases
Courtesy Toshiba
Coronary Angiography vs CT Angiography
Right coronary
artery showing
calcification
Spatial Resolution
CT angiography
Coronary Angiogram
Volume rendered CT angiogram of right
coronary artery acquired at 16x0.75 mm and
0.42 sec rotation time
Hoffmann et.al., AJR: 182, March 2004
Factors affecting Spatial Resolution
•
•
•
•
•
•
SDCT versus MDCT
MDCT detector array designs
Section thickness/Section collimation/
Effective section thickness
X-ray Tube
Tube Collimator
Collimated Slice
Pitch
Reconstruction Increment
Detector
Collimator
Reconstruction algorithms
Patient motion …
1-Row Detector
Single row detector CT
(SDCT)
8-Row Detector
Multiple row detector CT
(MDCT)
*Mahesh M, RadioGraphics, 22: 949-962, 2002
Mahadevappa Mahesh, MS, Ph.D.
Johns Hopkins University, Baltimore, MD
4
Detector Element Arrays* in 4-section MDCT scanners
16 x 1.25 mm
How are detector elements used in MDCT?
Uniform
20 mm
4 x 1.25 mm
Detectors
5
2.5 1.5 1 1 1.5 2.5
5
Non-uniform
20 mm
20 mm
4 x 0.5
Hybrid
15 mm
Switching Array
4-section scanners collect
4 simultaneous channels of data
15 mm
32 mm
Z-axis
*Mahesh M, RadioGraphics, 22: 949-962, 2002
Detector elements in 16 section MDCT scanners*
4 x 1.25 mm
Detector elements in >32 section MDCT scanners
4 x 1.25 mm
4 x 1.2 mm
32 x 0.6 mm
4 x 1.2 mm
16 x 0.625 mm
Siemens - Sensation 64
GE - Lightspeed 16
28.8 mm
20 mm
4 x 1.5 mm
16 x 0.75 mm
64 x 0.5 mm
4 x 1.5 mm
Toshiba - Aquilion 64
Siemens - Sensation 16
Philips - Mx8000 IDT
32 mm
6 x 1.25 mm
24 mm
40 x 0.6 mm
6 x 1.25 mm
Philips - Brilliance 40
16 x 0.5 mm
12 x 1 mm
40 mm
12 x 1 mm
64 x 0.625 mm
Toshiba - Aquilion 16
GE - Lightspeed 64
32 mm
40 mm
*Mahesh M, Clini Cardio Vasc Img Textbook, pp 1-77, 2004
Pitch redefined for MDCT
Beam Pitch =
I
Pitch: Definition, Confusion…
Detector Pitch =
W
T
Beam Pitch =
Detector Pitch
I - Table feed (mm/rotation)
W - Beam width (mm)
N
I
W
I
T
=
I
N*T
= Pitch†
T - Single DAS channel width (mm)
N - Number of active DAS channels
† IEC Part 2-44, 2003
Mahadevappa Mahesh, MS, Ph.D.
Johns Hopkins University, Baltimore, MD
5
Dose in MDCT varies as:
Why Cardiac CT protocols use Low Pitch†?
Z-position
• Pitch >1 implies extended
imaging and reduced patient
dose with lower axial
Helical
scan
direction
•
•
data overlap
Z-position
• Pitch <1 implies overlapping
and higher patient dose with
higher axial resolution
1
Pitch†
•
(mAs/rotation)
Time
Slope: Table feed speed
•
N - Number of active DAS channels
TR - Gantry Rotation Time (ms)
TRR - Time for one heart beat (ms)
M - Number of subsequent heart cycles
−
For ex: For heart rates 45-100 bpm, with TR 0.5 s,
TQ 250-360 ms, P = 0.375 to 0.875
−
P = 1.48
CTDI = 20.6 mGy
45% lower
Mahadevappa Mahesh, MS, Ph.D.
Johns Hopkins University, Baltimore, MD
For ex: For heart rate 60 bpm, with TR 0.4 s, N = 16 and
M = 2, P = 0.21
For ex: For heart rate 60 bpm, with TR 0.4 s, N = 16 and
M = 3, P = 0.15
Section Collimation (SC) vs Section Width (SW)
Effect of Pitch on Dose and Image Quality
P = 0.83
CTDI = 37 mGy
Data gaps
with higher pitch
Pitch is further restricted by the number of
segments used in reconstruction
N - Number of active DAS channels
TR - Gantry Rotation Time (ms)
TRR - Time for one heart beat (ms)
TQ - Partial scan rotation time (ms)
P = 0.64
CTDI = 47.8 mGy
30% higher
Typical pitch: 0.20 - 0.4
Multiple Segment Reconstruction
Retrospective ECG-gating with single segment
(partial scan) reconstruction requires limiting the
pitch dependent on heart rate
−
Hence pitch is low, and
radiation dose is high
•
Single Segment Reconstruction
•
High quality 3D with
minimal artifacts requires
resolution
Dose ∝
Higher pitch produces gaps
•
Section collimation is total beam
collimation divided by number of active
detector channels
•
Section width is the true thickness of
reconstructed image, measured as
FWHM of slice sensitivity profile
•
SW has to be larger than or equal to SC
•
SW ≈ 1.3 (±0.2) SC
Affected by collimation, pitch,
reconstruction algorithm, z-filter …
Slice Sensitivity Profiles:
conventional and spiral
acquisition
6
Reconstruction Interval (RI)
•
•
•
•
Defines degree of overlap between axial scans
Reconstruction Interval
•
Overlapping results in large number of images but
ensures optimum lesion display and improves MPR
and 3D images without increasing patient dose
For routine applications including small structures
detection a 30% section overlap is sufficient
•
•
For MPR and 3Ds, at least 50% overlap is desirable
Large RI yields fewer images but provides suboptimum lesion detection
•
Limited by reconstruction time, number of images
to interpret and storage space
Independent of section collimation or section width
Effect of Reconstruction Interval
Theoretic optimum is smaller than half the section
width with minimal added value in clinical practice
Effect of Reconstruction Interval
SW 0.5 mm, RI 0.3 mm
SW 0.5 mm, RI 5.0 mm
SW 0.5 mm, RI 0.3 mm
SW 0.5 mm, RI 0.5 mm
301 images
19 images
301 images
184 images
Effect of Reconstruction Algorthims:
Spatial Resolution and Image Noise Trade-offs
Effect of Slice Thickness on Image Noise
n
lutio
eso
lR
atia
Sp
10 mm
Smooth
Medium
5.0 mm
0.5 mm
Sharp
Reconstruction Filters
Smoothing filters results in minor reductions in spatial
resolution but require less dose for constant image noise
©Johns
Mahadevappa Mahesh, MS, Ph.D.
Johns Hopkins University, Baltimore, MD
Scan at thin collimation (raw data) for high spatial
resolution, but reconstruct thick sections with
overlaps to improve image noise
Hopkins 32 slice MDCT
7
High Contrast Spatial Resolution in Z direction
• Phantom
- 0.5 mm spacing
• Scan modes
- 32 x 0.5 mm
• Technique
- 120 kVp, 50 mAs
Spatial Resolution
•
Axial or In-plane Resolution
−
•
MDCT: 10 - 20 lp/cm (resolvable ~ 0.5- 0.25 mm)
Z-axis Resolution - influenced by section
collimation
0.6 mm
0.5 mm
0.4 mm 0.3 mm
©Johns
−
MDCT: 7- 15 lp/cm (resolvable ~ 0.7 - 0.3 mm)
Hopkins
CT Dose Index (CTDI) comparison:
4 vs 16* vs 64 MDCT¶ scanner
Geometric Efficiency
Sensation 64: Increase in CTDIw of nearly 4-19% compared to
Sensation 16 as measured for head and body phantoms
¶Siemens
* CTDIw(weighted average) normalized to 16x0.75 mm scan mode
Geometric Efficiency
Focal spot
•
Amount of radiation excluded (penumbra)
Geometric Efficiency
Beam
collimator
•
Decreases with thinner sections and fewer
detector elements
Penumbra
•
Improves with thicker sections or with more
active thin sections
relative to the radiation collected by the
detectors in forming an image
•
SDCT
Penumbra caused by finite focal spot size
−
contributes to image and patient dose in SDCT
−
contributes to patient dose but not to image in MDCT
Penumbra
Penumbra
Penumbra
(over-beaming to ensure equal image quality)
MDCT
Mahadevappa Mahesh, MS, Ph.D.
Johns Hopkins University, Baltimore, MD
Penumbra
4 x 1.25 mm
4 x 2.5 mm
16 x 1.25 mm
Strong effect with
thin collimation
Effect decreases with
thicker collimation
Effect diminishes further
with more detectors
8
Advantage of Thin Sections
Geometric Efficiency in MDCT
•
Dose in 4 slice scanners
grows markedly with thin
collimation but less so
for 16 and 16+ scanners
In general, thin sections yields higher z-axis
resolution, improves partial volume, but requires
higher tube current to reduce image noise, which
leads to higher doses (especially with few detectors)
Dose grows markedly with
thin collimation and few
active detector elements but
less so for thicker collimation
and more detectors
©Johns
Hopkins
©Johns
Hopkins MDCT scanners
Effect of X-ray Beam Energy (kVp)
Effect of kV and mAs on Image Noise
135 kV with 8.0 SD
37 mGy CTDI
120 kV with 10.0 SD
29 mGy CTDI
~22% dose reduction
Effect of kV and mAs on Image Noise
Artifacts
Mahadevappa Mahesh, MS, Ph.D.
Johns Hopkins University, Baltimore, MD
9
Pulsation Artifacts
•
Common artifacts due
to cardiac pulsation
•
Multiple segment
reconstruction is
desirable
Banding Artifacts
Banding artifacts
Artifacts due to increased heart rate
starting at 51 bpm increased to 69 bpm
Radiographics 2005
Radiographics 2005
Artifacts due to incomplete breath holding
Streak Artifacts
Coronal
Sagittal
Metallis structure but
no artifact visible on
axial images
Streak artifacts due to earlier stent placement
visible on thin MIP and MPR images
Axial images show no artifacts
Radiographics 2005
Impact of Dose Modulation: Chest CT
MDCT 64*
CT Dose Modulation
Radiation dose:
Lateral: 16% increase, AP: 25% reduction
*Mahesh, Kamel & Fishman, Evaluation of ‘CareDose’ on Siemens Sensation 64 MDCT scanner
Mahadevappa Mahesh, MS, Ph.D.
Johns Hopkins University, Baltimore, MD
10
Impact of Dose Modulation: Abdominal CT
MDCT 64*
Radiation dose:
Lateral: 28% increase, AP: 48% reduction
*Mahesh, Kamel & Fishman, Evaluation of ‘CareDose’ on Siemens Sensation 64 MDCT scanner
Conclusions
Dose Modulation for varying Image Noise Index
•
Cardiac imaging is highly demanding application of
MDCT and is possible due to technological advances
•
Understanding trade-offs between various scan
parameters that affects image quality is key
•
Low SD - High dose
184 mA
Medium SD & dose
69 mA
High SD - Low dose
48 mA
Cardiac CT has the potential to becoming reliable
tool for noninvasive diagnosis and prevention of
cardiac and coronary artery disease
Toshiba
Future with CT
Mahadevappa Mahesh, MS, Ph.D.
Johns Hopkins University, Baltimore, MD
11