Pitfalls and Remedies of MDCT Scanners as
Quantitative Instruments
Root-Causes of CT Number Inaccuracies
Pitfalls and Remedies of MDCT Scanners as
Quantitative Instruments
Jiang Hsieh, PhD
GE Healthcare Technology
University of Wisconsin-Madison
•
Nature of the X-ray Physics
•
New Technology
scanner
– Beam Hardening
– Scatter
– Aliasing
– Helical
– Cone Beam
– Incomplete Sampling
•
Patient
•
Operator
operator
– Motion
– Photon Starvation
– Protocols (scan thin, recon thick)
– Partial Volume
patient
1
2
Nature of Physics: Beam Hardening
measured
poly-energetic x-ray photon
energy-dependent m(E)
measured projection error
polynomial correction
N
p'

n p
n
n 1
channel
0 cm water
•
The attenuation characteristics of bone is
significantly different from that of soft tissue.
Similar observation can be made for contrast
agents.
1000
m (1/cm)
•
•
•
•
•
m(E)
intensity
ideal
10000
Bone Beam Hardening
100
bone
10
water
1
0.1
0
50
100
150
energy (keV)
energy, E
15 cm water
30 cm water
no correction
with correction
3
Water BH
Error
Bone BH
4
1
Pitfalls and Remedies of MDCT Scanners as
Quantitative Instruments
Dual Energy Correction
Impact of Iodine Contrast
• Dual energy takes advantage of the two types of x-ray interaction with
matter: photoelectric and Compton.
• Use of the additional information can effectively remove BH.
• Iodine contrast
– Cardiac phantom experiments
– Dual energy solution
Conventional 120kVp
Dual Energy 70keV
100000
10000
m (1/cm)
1000
Iodine
100
10
Bone
1
Soft-tissue
0.1
0
80 kVp
140 kVp
Perfusion
100
150
Cardiac Phantom
energy (keV)
Projection-space DECT
ww=300
Regular 140 kVp
50
5
6
DE Mono 70 keV
Metal-induced Error
• Perfusion is important to test
the viability of a tissue
•
• Perfusion requires accurate
quantitation of CT numbers
•
• BH is one of the major
sources of error
CT is sensitive to the presence of high-density and highfrequency objects.
Special care must be taken to avoid placing foreign objects
inside the scanner FOV.
0.625mm thickness, 1.0mm apart
140kVp
Images courtesy of Drs. T. Lee
and A. So of St. Mary’s
Hospital, London, OT, Canada
metal reinforcement
7
8
2
Pitfalls and Remedies of MDCT Scanners as
Quantitative Instruments
Nature of Physics: Scatter
•
•
Scatter-to-Primary Ratio (SPR) increases almost
linearly with the z-coverage.
SPR decreases with the increase in kVp
SPR increases with the object size
•
•
X-ray source
Scatter Artifacts
•
•
Scatter-to-Primary Ratio (SPR) increases almost linearly
with the z-coverage.
SPR decreases with the increase in kVp
SPR increases with the object size
z
scatter
scatter
M e a s u re d S P R in V C T /H D a t 4 0 m m c o v e ra g e
z
ghost image
0 .5
120kVp 48cm Poly
0 .4
30
25
0 .2
SPR
8 0 kVp
1 0 0 kVp
0 .1
1 2 0 kVp
20
15
10
0
shading
5
15
25
35
45
55
w a te r d ia m e te r (c m )
0
0
50
100
Beam Width (mm)
150
200
9
Lung CT Number Accuracy
Inspiration Level
Beam Hardening
– Increased cone beam effect
– Increased scatter
-980
Sagittal view
-985
32mm Helical
-990
Scatter
– Increased volume coverage
– Insufficient post-patient
collimation
10
• Air CT number can be impacted by
– Patient centering
– Variation of object size
– Presence of non-water materials
•
160mm Detector
Air-Lung Accuracy
Lung Volume
•
•
32mm Detector
-995
Axial
CT Number
CT Number (HU)
S PR
35
0 .3
-1000
-1005
-1010
-1015
160mm Axial
-1020
JP Sieren, PF Judy, DA Lynch, JD Newell, HO Coxson, and EA
Hoffman, “QIBA Quantitative CT: Towards routine quantitative CT
in obstructive lung disease,” QIBA COPD/Asthma Subcommittee
-1025
-1030
1
11
51
101
151
201
12
3
Pitfalls and Remedies of MDCT Scanners as
Quantitative Instruments
Truncation Issue
Scatter Correction
•
•
•
•
• For x-ray radiography, truncation does not cause artifact
inside FOV.
• Truncated projection will cause significant artifacts for CT
Increased collimator aspect ratio: H/W
2D collimation
Algorithmic correction possible
Phantom calibration
COPDGene Phantom (CTP657)
W
Chest X-ray
H
z
x
Full FOV
original
14
Size Matters
•
CT is sensitive to both high density object and
objects extended outside the FOV.
•
Patient arms need to be positioned outside the FOV.
Truncated
corrected 13
Photon Starvation
50cm FOV
•
Beer’s law indicate that the amount
of attenuation increases
exponentially with path length.
I
e
 mL
I0
•
•
Scout Image
CT Image
15
At low signal level, the noise in the
projection is no longer dominated by
the x-ray photon.
Convolution filtering operation will
further amplify the noise and streak
artifacts will result.
16
4
Pitfalls and Remedies of MDCT Scanners as
Quantitative Instruments
Protocol Optimization
Protocol optimization
–
–
–
–
–
–
Scan speed
Helical pitch
kVp selection
Slice thickness
Reconstruction kernel
mA modulation
m (1/cm)
•
•
Adaptive filtering for streak reduction
10
Bone
1
soft-tissue
0.1
0
30
60
90
120
150
energy (keV)
80kVp, 6.4mGy
adaptively filtered
140kVp, 6.3mGy
17
Iterative Reconstruction
120 kV, 10 mA, 0.5 rot time
18
Sinogram
Cardiac Motion  16-slice Examples
Dose report
FBP
original
Algorithmic Correction
100
MBIR
0.625mm slice thickness
•
•
•
Complexity of cardiac motion
Artifacts from several factors
In-plane vs. cross-plane temporal resolution
.78mSv*
phase mis-registration
sub-optimal gating
Images courtesy of Dr Barrau, CCN, France
*Obtained by EUR-16262 EN, using an abdomen factor of 0.015*DLP and a pelvis factor of 0.019*DLP
In-Plane
sub-optimal table speed
Cross-Plane
20
5
Pitfalls and Remedies of MDCT Scanners as
Quantitative Instruments
Artifact Correction
•
Advanced Technology-induced Artifact
In-Plane Temporal Resolution Improvement:
–
–
–
–
•
•
•
•
•
Phase optimization
Faster scan speed
Multiple source-detector
Advanced reconstruction
Half-scan: p + fan angle acquisition.
Dual source reduces the acquisition by 40-45%
Two detectors have different size
Object outside small FOV may cause artifacts.
Compensation algorithm can reduce the impact
centered phantom
half-scan
sub-optimal phase
•
•
•
•
optimal phase
25 g at 0.35 s
8X safety margin  200 g
76 g at 0.2 s
8X safety margin  612 g
•
•
Cone Beam Axial
Axial Truncation
desired ROI
detector
ww=600
z
source trajectory
phase mis-registration
22
• Exact cone-beam reconstruction does not exist.
• Large cone angle can lead to cone-beam artifact.
Increase z-coverage will reduce phase
mis-registration artifacts.
Entire heart covered by a single rotation.
Step-and-shoot mode artifacts are due to:
– Missing frequencies
– Mis-handled frequencies
– Z-truncation
smaller
detector FOV
21
Cross-Plane Temporal Resolution:
Cone Beam CT
•
off-centered phantom
23
Small Detector
Helical Mode
Large Detector
Step-and-shoot
24
6
Cone Beam Axial
Temporal Resolution Improvement:
Axial, Phantom
Advanced Reconstruction Algorithms
• Exact cone-beam reconstruction does not exist for
step-and-shoot mode.
• Large cone angle can lead to cone-beam artifact.
• Active research in on-going
Characterize cardiac motion
Compensate for motion
Coronal, Simulation
Advanced
Axial, Phantom
•
•
Conventional
Pitfalls and Remedies of MDCT Scanners as
Quantitative Instruments
Conventional
Rest/Low HR
25
Conventional
Stress/High HR
Advanced
Stress/High HR
High HR
+ motion
correction
26
Conclusion
• There are many sources of error for quantitative CT
–
–
–
–
Nature of Physics
New Technologies
Patient
Operator
• Many approaches are taken to reduce or eliminate the error.
• It takes a combined efforts from manufactures, CT
operators, and patient corporation to reduce the impact of these
pitfalls to a minimum.
27
28
7