Professional Documents
Culture Documents
Relating Noise To Image Quality Indicators in CT Examinations With Tube Current Modulation
Relating Noise To Image Quality Indicators in CT Examinations With Tube Current Modulation
Solomon et al.
Noise and Image Quality Indicators in CT Using Tube Current
Modulation
A
dvances in x-ray CT technology noise according to the selected image quality
and realization of its unprecedent- indicator. Therefore, an image quality indica-
ed clinical utility have resulted in tor can be thought of as a surrogate for expect-
an increase in the number of CT ed noise. The ATCM system eliminates the
examinations performed in the United States. need for a technologist to manually adjust the
This has resulted in a large increase in radiation tube current for different patients while at the
Keywords: CT, noise, noise index, reference tube
dose received by patients as a result of CT ex- same time automatically adjusting the tube
current, tube current modulation aminations [1, 2]. Automatic tube current mod- current to accommodate differing attenuation
ulation (ATCM) techniques were developed, properties within a patient habitus.
DOI:10.2214/AJR.12.8580 in part, in an effort to reduce dose and main- When designing and optimizing CT proto-
tain image quality for patients of different siz- cols, one is required to balance the need for
Received January 17, 2012; accepted after revision
June 4, 2012. es and shapes [3–6]. In CT examinations using acceptable image quality with the mandate to
ATCM, a level of image quality is targeted by keep radiation exposure “as low as reason-
1
All authors: Carl E. Ravin Advanced Imaging Laborato- setting an image quality indicator. The ATCM ably achievable.” This is especially important
ries and Clinical Imaging Physics Group, Department of systems then use patient size and attenuation for clinical tasks, such as pediatric lung nod-
Radiology, Medical Physics Graduate Program, Duke
properties to modulate the x-ray tube current ule detection, that are particularly sensitive to
University Medical Center, Hock Plaza, 2424 Erwin Rd,
Ste 302, Durham, NC 27705. Address correspondence to (i.e., photon fluence) to create images of the image quality. Li et al. [8] have shown that it
J. B. Solomon (justin.solomon@duke.edu). desired quality [7]. Generally, modulating the is possible to establish a relationship between
tube current according to patient size and shape measureable image quality and radiologist per-
AJR 2013; 200:592–600
is sought to control the photon fluence at the formance in pediatric lung nodule detection.
0361–803X/13/2003–592 detectors. Because image noise is inversely ATCM makes it possible to design protocols
proportional to the square root of fluence, these that will produce images of similar quality for
© American Roentgen Ray Society systems are attempting to control quantum patients of different shapes and sizes [9], but
what measureable noise levels can be expect- according to the Q and individual patient size ferent patient categories was further explored on
ed for a given quality reference indicator? In and attenuation properties [9]. The Siemens the Somatom Definition Flash.
other words, if we know what noise level is Healthcare ATCM algorithm does not alter the Because of scan time restraints and the num-
acceptable to achieve a certain level of di- tube current according to any reconstruction ber of parameters investigated, data were not col-
Downloaded from www.ajronline.org by Duke University Medical Center on 03/03/13 from IP address 152.16.51.172. Copyright ARRS. For personal use only; all rights reserved
agnostic performance, how can we choose a parameters, nor does it attempt to achieve con- lected for every combination of the parameters
quality reference indicator to produce imag- stant noise levels across patients of different listed. Also, system limitations on tube output
es with the desired noise level? Complicating sizes. Given such a design and known physics made some acquisitions unachievable. For exam-
this answer is the fact that different manufac- of CT, one would expect actual image noise ple, at 80 kVp, the Discovery CT 750 HD cannot
turers apply ATCM differently and use dif- to be inversely related to the Q and be depen- achieve a tube current sufficient to image an adult
ferent quality reference indicators. dent on reconstruction parameters and patient patient at an NI of 6. In general, data were collect-
The purpose of this study was to quantita- size. The purpose of the study was to investi- ed by first choosing a reference protocol accord-
tively assess the relationship between actual gate how the actual noise in a patient image ing to clinical settings. The phantom was imaged
quantum noise and image quality indicators varies as a function of NI and Q across a range using this reference protocol at various values of
across imaging parameters in the context of of imaging parameters. the system’s image quality indicator. The refer-
two specific commercial CT platforms man- ence protocol was then changed one parameter
ufactured by GE Healthcare and Siemens Materials and Methods at a time, each time imaging with several image
Healthcare. Both platforms use simultaneous A 64-MDCT scanner (Discovery CT 750 HD, quality indicator values. This made it possible to
longitudinal (z-axis) and angular (x–y) mod- GE Healthcare) and a 128-MDCT scanner (So- systematically explore how each parameter af-
ulation, which has been shown to decrease matom Definition Flash, Siemens Healthcare) fects the relationship between noise and the im-
dose compared with systems that use only an- were chosen for this study. We investigated the age quality indicator in question. Reference pa-
gular modulation or no modulation [10]. On ATCM algorithm SmartmA (GE Healthcare) on rameters are indicated with footnotes in Table 1.
the GE Healthcare scanner, one selects a noise the Discovery CT 750 HD scanner and the ATCM Note that, on the Discovery CT 750 HD, the mini-
index (NI) and specifies minimum and maxi- algorithm CARE Dose 4D (Siemens Healthcare) mum and maximum tube current were set to their
mum tube current levels. NI is expected to re- on the Somatom Definition Flash scanner. Two lowest and highest available values to ensure that
flect the noise in the central region of a uni- anthropomorphic phantoms, representing an adult the required tube current (as determined by the
form phantom. The GE Healthcare ATCM male (73 kg) and a 1-year-old child (10 kg) (ATOM SmartmA algorithm) was not clipped during the
algorithm will attempt to produce images with models 701 and 704, Computerized Imaging Ref- scan. The phantoms were scanned from the top of
noise consistent with the selected NI, indepen- erence Systems), were imaged on the two systems. the lungs through the abdomen to include a range
dent of peak kilovoltage, patient size, and pro- A clinical range of imaging parameters was of anatomic and attenuation variations. Each scan
spectively chosen reconstructed slice thick- chosen to investigate the relationship between im- was repeated yielding duplicate series. Duplicates
ness [7]; therefore, one would expect actual age quality indicators and measured noise. The were then subtracted from each other to remove
image noise to correlate positively with NI parameters of interest were beam collimation, he- anatomy (i.e., signal) (Fig. 1).
and be independent of peak kilovoltage, pa- lical pitch factor, peak tube voltage, slice thick- For each series, three square regions of interest
tient size, and prospective slice thickness. On ness, reconstructed FOV, reconstruction algo- (ROIs) were defined: one in the mediastinum, one in
a Siemens Healthcare scanner, one selects an rithm, and the convolution kernel (reconstruction the lung, and one in the abdomen. Noise magnitude
image quality reference effective tube current- filter) (Table 1). Because Siemens Healthcare’s (pixel SD) in the subtracted images was measured
time product (Q). The Q represents the mean CARE Dose 4D modulation uses different atten- over multiple slices and averaged. These noise mea-
effective tube current used on a reference pa- uation curves for different patient age categories surements were divided by the square root of 2 to ac-
tient. The system then adapts the tube current (e.g., adult or pediatric), the effect of selecting dif- count for the noise added as a result of the subtraction
changing the ROI size or location within the ing this model, the average residual magnitude
14 was found to be 0.46 HU (5% error) with a
anatomic structure of interest was found to
12 have little effect on the noise measurement. maximum at 2.5 HU (18% error). The root-
10 Overall, the variations due to ROI size (from mean square of residuals was 0.7 HU. In gen-
8 400 to 1600 pixels) and location (variable eral, residual magnitude was larger for noisier
6 within the anatomic region) were 3% and images, whereas the less noisy images resulted
4
8%, respectively. Second, the SD associat- in a higher percentage error. Please note that
4 9 14 19 ed with each noise measurement increased as the adjustment factors listed in Table 2 pertain
Noise Index the average noise increased. The average SD only to the range of values studied for a given
was 0.6 HU, and the maximum was 2.6 HU. parameter. The factors may not be directly ex-
Fig. 2—Plot of noise in mediastinum against noise
index for adult phantom under reference conditions
The average coefficient of variation was 9%, trapolated to other parameter values.
(120 kVp, pitch of 1.375, 40-mm beam collimation, and the maximum was 19%. On the Somatom Definition Flash, for every
slice thickness of 5 mm, and reconstruction with On the Discovery CT 750 HD, for every set of scan parameters, noise decreased with in-
34-cm FOV using filtered back projection and set of scan parameters, noise increased lin- creasing image Q. An example is given in Fig-
standard filter). Dashed line is added as reference to
show where noise index is equal to noise. Error bars early with NI. An example is illustrated in ure 5 for the reference protocol. The noise-Q
represent 1 SD. Figure 2 for the reference protocol. At a giv- relationship had little dependence on helical
A B C
25 Coefficient of 14 Coefficient of 12 Difference = 3%
variation = 3% variation = 7%
12
Downloaded from www.ajronline.org by Duke University Medical Center on 03/03/13 from IP address 152.16.51.172. Copyright ARRS. For personal use only; all rights reserved
20 10
10
Coefficient of 8
Noise (HU)
Noise (HU)
Noise (HU)
Coefficient of
15 variation = 3% 8
variation = 8% Difference = 2%
6
10 6
4
4
5 2
2
0 0 0
11.57 22.00 6.0 11.57 6.0 11.57
Noise Index Noise Index Noise Index
Slice thickness = 0.625 mm Slice thickness = 1.25 mm kVp = 80 kVp = 100 Pitch = 0.984 Pitch = 1.375
Slice thickness = 5.00 mm kVp = 120 kVp = 140
D E F
12 Difference = 6% 14 12 Difference = 6%
Difference = 4%
10 12 10
10
8 Difference = 1% 8
Noise (HU)
Noise (HU)
Noise (HU)
8 Difference = 3%
Difference = 4%
6 6
6
4 4
4
2 2 2
0 0 0
6.0 11.57 6.0 11.57 6.0 11.57
Noise Index Noise Index Noise Index
FOV = 34 cm FOV = 40 cm Beam width = 40 mm Beam width = 20 mm Adult One-Year-Old
G H
35 Coefficient of 14 Difference = 21%
variation = 64%
30 10
25
Coefficient of 8
Noise (HU)
Noise (HU)
pitch factor (< 6% difference), FOV (< 5% dif- age, for a given image series, lung noise was where σ, σref, and A are the same as in Equation
ference), or detector configuration (< 8% dif- 27% lower and abdomen noise was 24% higher 1; e is Euler’s number, α, β, and γ are fitting
ference) (Fig. 6). However, the relationship was compared with mediastinum noise measured in parameters; Q is the quality reference effective
dependent on slice thickness, peak kilovoltage, the same image series. This is illustrated by the tube current; t is the slice thickness in millime-
phantom size, patient age selection, anatomy plots in Figure 7. ters; and kVp is the peak kilovoltage of the x-
of interest, and reconstruction algorithm (Figs. Given the design of the CARE Dose4D al- ray tube. The fitting parameters α, β, and γ
6 and 7). The significant impact on the noise- gorithm [11], and the results already present- were found using a linear least-squares fit to
Q relationship due to slice thickness, peak ki- ed, the noise-Q relationship is best described the natural log of noise. The adjustment fac-
lovoltage, phantom size, age selection, anato- by Equation 3: tor, A, was found by taking ratios of noise un-
my of interest, or reconstruction algorithm was eα der conditions of interest to noise under refer-
found to be significant according to the criteria σ = A • σref = A • (3), ence conditions and describes the dependency
(Q • t)β • kVpγ
described in Materials and Methods. On aver- of the noise-Q relationship on the anatomy of
40 y = 0.694x 30 y = 1.254x for the same phantom and scan settings, ex-
R2 = 0.99143 R2 = 0.99602 pected noise may be different depending on
35
25 the selection of an adult or pediatric patient.
Lung 0.69 6
Abdomen 1.25 4
Convolution kernel 2
Standard 1.00 0
0 50 100 150 200 250 300 350
Bone 2.62
Q (mAs)
Soft 0.81
Fig. 5—Plot of noise in mediastinum against image
Detail 1.05 quality reference effective tube current-time product
Reconstruction algorithm (Q) for adult phantom under reference conditions
(120 kVp, pitch of 0.8, 32.8-mm beam collimation
Filtered back projection 1.00 with flying focal spot, 5-mm slice thickness, and
reconstruction with 34-cm FOV using filtered back
30% adaptive statistical iterative reconstruction 0.84
projection and B31f filter). Error bars represent 1 SD.
A B C
60 35 Coefficient 18
Coefficient of Coefficient Difference = 5%
of variation = 40% of
variation = 55%
30 variation = 49% 16
Downloaded from www.ajronline.org by Duke University Medical Center on 03/03/13 from IP address 152.16.51.172. Copyright ARRS. For personal use only; all rights reserved
50
Coefficient 14
of 25 Difference = 1%
40 variation = 55% 12
Noise (HU)
Noise (HU)
Noise (HU)
Coefficient
of 20 10
30 variation = 55% Coefficient
15 of
variation = 53%
8
20 6
10
4
10 5
2
0 0 0
75 150 300 75 150 300 75 150
Q (mAs) Q (mAs) Q (mAs)
Slice thickness = 0.6 mm Slice thickness = 1.0 mm kVp = 80 kVp = 100 Pitch = 0.8 Pitch = 1.4
Slice thickness = 3.0 mm Slice thickness = 5.0 mm kVp = 120 kVp = 140
D E F
18 7 Difference = 7% 16 Difference = 59%
Difference < 1%
16 6 14
14 Difference < 1% 12 Difference = 57%
Difference = 2%
5
12
Noise (HU)*
Noise (HU)
Noise (HU)
10
10 4 Difference = 2% Difference = 58%
Difference = 4% 8
8 3
6
6
2
4 4
2 1 2
0 0 0
75 150 300 30 60 120 75 150 300
Q (mAs) Q (mAs) Q (mAs)
FOV = 34 cm FOV = 40 cm 128 × 0.6 32 × 1.2 Adult One-Year-Old
G H
8 18
Difference = 37% Difference = 35%
7 16
6 14
12 Difference = 36%
Noise (HU)*
Difference = 33%
Noise (HU)
5
10 Fig. 6—Noise in mediastinum against image quality
Difference = 36%
4 Difference = 30% reference effective tube current-time product (Q).
8 A–H, Bar graphs show effect of slice thickness (a),
3
6 peak kilovoltage (b), pitch (c), FOV (d), detector
2 4 configuration (e), phantom size (f), patient age
1 selection (g), and reconstruction algorithm (H)
2
on relationship between noise and Q. Each plot
0 0 represents alteration with respect to reference
75 150 300 75 150 300
technique by one parameter. Reference parameters
Q (mAs) Q (mAs)
are underlined in key of each plot. Asterisks indicate
Adult selected Pediatric selected Filtered back projection that these noise values were measured on 1-year-old
Iterative reconstruction in image space phantom. Error bars represent 1 SD.
scanners from multiple manufacturers. Us- age dependence of the noise-NI relationship physics principles (i.e., the noise is roughly in-
ing Equations 1–4, protocols could be written is not consistent with the design of SmartmA versely proportional to the square root of slice
to achieve similar noise levels from different and will be discussed further later in this arti- thickness) [12].
scanners for patients who are similar in size to cle. Measurable noise in different ROIs can be As mentioned already, on the Discovery
the phantoms used in this study. predicted from the selection of NI. One caveat CT 750 HD, changing the peak kilovoltage
Results on the Discovery CT 750 HD are is that the SmartmA algorithm will modulate did affect the noise-NI relationship, but as
mostly consistent with the design of SmartmA the tube current according to the prospective- Table 2 suggests, that effect was small com-
modulation [7]—that is, the system was able ly chosen reconstructed slice thickness. This pared with other factors. One must keep in
to modulate the tube current to achieve im- implies that changing the slice thickness pro- mind that noise magnitude is only one of sev-
ages of similar noise for different slice thick- spectively could have a large effect on patient eral metrics of image quality. This means that
nesses, pitches, and phantom sizes. Changing exposure. A retrospective reconstruction for a two images with identical noise magnitude
peak kilovoltage, however, did slightly affect different slice thickness will result in a change acquired at different peak kilovoltages do not
the noise-NI relationship. The peak kilovolt- in noise magnitude as predicted by known necessarily have the same image quality. For
40 y = 0.7322x 70 y = 1.2362x
noise-Q relationship. For this reason, Equation
R2 = 0.99296 R2 = 0.99508 1 (representing SmartmA) has no dependence
35 60 on slice thickness, whereas Equation 3 (rep-
50
Lung Noise (HU)
TABLE 3: Fitting Constants and Adjustment Factors Needed to Calculate voltage. This function can be further adjust-
Expected Noise for a Given Protocol Using Equations 2 and 4 ed to account for differing anatomy, phantom
Protocol Adjustment Factor size, patient age selection, and reconstruc-
tion algorithm. This function is independent
Downloaded from www.ajronline.org by Duke University Medical Center on 03/03/13 from IP address 152.16.51.172. Copyright ARRS. For personal use only; all rights reserved
2nd ed. Philadelphia, PA: Lippincott Williams & dose in temporal bone high-resolution multidetec- struction algorithm. AJR 2010; 194:191–199
Wilkins, 2001:367 tor row CT. AJNR 2011; 32:1783–1788 21. Leipsic J, Labounty TM, Heilbron B, et al. Adap-
13. Bogot NR, Fingerle A, Shaham D, Nissenbaum I, 17. Glover GH. Compton scatter effects in CT recon- tive statistical iterative reconstruction: assessment
Sosna J. Image quality of low-energy pulmonary structions. Med Phys 1982; 9:860–867 of image noise and image quality in coronary CT
Downloaded from www.ajronline.org by Duke University Medical Center on 03/03/13 from IP address 152.16.51.172. Copyright ARRS. For personal use only; all rights reserved
CT angiography: comparison with standard CT. 18. Boedeker KL, McNitt-Gray MF. Application of angiography. AJR 2010; 195:649–654
AJR 2011; 197:384; [web]W273–W278 the noise power spectrum in modern diagnostic 22. Ghetti C, Ortenzia O, Serreli G. CT iterative re-
14. Flohr TG, Schaller S, Stierstorfer K, Bruder H, MDCT. Part II. Noise power spectra and signal to construction in image space: a phantom study.
Ohnesorge BM, Schoepf UJ. Multi-detector row noise. Phys Med Biol 2007; 52:4047–4061 Phys Med 2012; 28:161–165
CT systems and image-reconstruction techniques. 19. Miéville FA, Gudinchet F, Rizzo E, et al. Paediat- 23. Lee CH, Goo JM, Ye HJ, et al. Radiation dose
Radiology 2005; 235:756–773 ric cardiac CT examinations: impact of the itera- modulation techniques in the multidetector CT
15. Nicholson R, Fetherston S. Primary radiation out- tive reconstruction method ASIR on image quali- era: from basics to practice. RadioGraphics 2008;
side the imaged volume of a multislice helical CT ty—preliminary findings. Pediatr Radiol 2011; 28:1451–1459
scan. Br J Radiol 2002; 75:518–522 41:1154–1164 24. Söderberg M, Gunnarsson M. The effect of differ-
16. Niu YT, Olszewski ME, Zhang YX, Liu YF, Xian 20. Silva AC, Lawder HJ, Hara A, Kujak J, Pavlicek ent adaptation strengths on image quality and ra-
JF, Wang ZC. Experimental study and optimiza- W. Innovations in CT dose reduction strategy: ap- diation dose using Siemens Care Dose 4D. Radiat
tion of scan parameters that influence radiation plication of the adaptive statistical iterative recon- Prot Dosimetry 2010; 139:173–179