Physics and Imaging in Radiation Oncology 4 (2017) 6–11

Contents lists available at ScienceDirect

Physics and Imaging in Radiation Oncology

journal homepage: www.elsevier.com/locate/phro

Original Research Article

Improving CT quality with optimized image parameters for radiation T

treatment planning and delivery guidance
⁎
Guang-Pei Chen , George Noid, An Tai, Feng Liu, Colleen Lawton, Beth Erickson, X. Allen Li
Department of Radiation Oncology, Medical College of Wisconsin, Milwaukee, WI 53226, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Background and purpose: CT scan protocols are often created with imaging parameters set to minimize imaging
IGRT dose with acceptable image quality for diagnostic purpose. This study aimed to optimize CT imaging parameters
CT image quality to help accurately delineate structures for radiation therapy planning and delivery guidance.
CT image dose Materials and methods: Imaging parameters were optimized with CT data acquired for a phantom to create image
quality enhancement (IQE) protocols, which were subsequently used to scan a prostate and a pancreatic cancer
patient who underwent image-guided radiotherapy (IGRT). The patient images were compared with those
scanned with standard clinical protocols, the quality of these images was assessed with various methods (survey,
inter- and intra-observer variations, and dice coefficient analysis) for the two patient cases.
Results: An effective tube current–time product of ∼1000 mAs was found to be a reasonable choice to balance
CT quality and CT dose. With increased dose and penetration taken into account, 100 and 120 kV tube voltages
were found appropriate for the IQE protocols. The inter- and intra-observer variations for the IQE data were
smaller than those with the standard protocols. Dice coefficient analysis indicated that the IQE protocols lead to
improved dice coefficient by as much as 8 percentage points for the two cases studied.
Conclusion: CT image quality can be improved with the IQE protocols created in this study, to provide better soft
tissue contrast, which would be beneficial for use in radiation therapy, e.g., for planning data acquisition or for
IGRT for hypo-fractionated treatments.

1. Introduction target volume (CTV), and to add another margin surrounding CTV to
create planning target volume. These margins are introduced to account
Computed tomography (CT) has been widely used in diagnostics for various factors, including structure delineation uncertainties. It is
and radiotherapy (RT). Dose from CT imaging has an associated risk highly desirable to reduce these margins for the purpose of reducing
and should be kept low [1]. For diagnostic purposes, CT dose is often radiation dose to organs at risk. On the other hand, IGRT may have
minimized as long as CT quality is acceptable [2]. For RT, CT has been limited clinical value due to lack of accurate structure delineation [4].
used mainly for RT planning (RTP) and, in recent years, for guiding RT Thus, improving CT quality to reduce structure delineation uncertainty
delivery, e.g., image guided radiotherapy (IGRT). High CT quality is in RTP and to reduce alignment error in IGRT is helpful and crucial for a
essential for accurate 3D structure delineation in both RTP and IGRT. successful treatment.
Consequently, goals for using CT in diagnostics and RT are different. As Therefore, in radiation oncology, an increase in CT dose in exchange
Hevezi pointed out, CT scan protocols should be adjusted to obtain for improved CT quality may be justified. The CT dose may be negli-
sufficient CT image rendition for the planning procedure at hand [3]. gible in comparison to the therapeutic dose. In particular, for some
This may require an actual increase in CT scan technique from that used special procedures, such as stereotactic body radiotherapy (SBRT), the
for reduced-dose diagnostic techniques. The CT acquisition protocols dose increase from CT scan, as compared to the prescription dose, can
that are often optimized for diagnostic radiology (often provided by be acceptable. Based on phantom studies, Li et al. developed a general
vendors) may not be optimal for RT. The ultimate goal of RT is to de- strategy to predict the optimal CT simulation protocols in a flexible and
liver a high dose of radiation to a tumor while sparing the surrounding quantitative way that takes into account patient size, treatment plan-
normal tissues as much as possible. In RTP, it is a common practice to ning task, and radiation dose [5]. They defined an image quality index
add a margin surrounding gross tumor volume (GTV) to form clinical (IQI) to act as a surrogate linking the optimal simulation protocol to the

⁎
Corresponding author.
E-mail address: gpchen@mcw.edu (G.-P. Chen).

http://dx.doi.org/10.1016/j.phro.2017.10.003
Received 15 December 2016; Received in revised form 25 October 2017; Accepted 26 October 2017
2405-6316/ © 2017 The Authors. Published by Elsevier Ireland Ltd on behalf of European Society of Radiotherapy & Oncology. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
G.-P. Chen et al. Physics and Imaging in Radiation Oncology 4 (2017) 6–11

contouring and treatment planning task. Their study indicated that the CT scanner has the Automatic Exposure Control capability, known as
optimal CT simulation protocol and the corresponding radiation dose CARE Dose4D [14]. With CARE Dose4D, the user sets adaptation
varied significantly for different patient sizes, contouring accuracy, and strength, selects a tube voltage (kV) setting and determines a quality
radiation treatment planning tasks. Furthermore, alternative CT mod- reference tube current–time product (mAs). A topogram is performed
alities with higher CT doses, such as megavoltage CT or cone beam CT prior to the actual scans to determine tube current values at different
have been commonly used in IGRT. By increasing the linear accelerator angular and axial positions so that patient size and attenuation changes
(LINAC) pulse rate, Westerly et al. studied high-dose imaging modes on can be adapted.
a clinical tomotherapy machine [6]. They concluded that increasing the
imaging dose results in increased contrast-to-noise ratio (CNR), and 2.2. Phantom measurements
makes it easier to distinguish the boundaries of low contrast objects.
The American association of physicists in medicine (AAPM) has re- To find a balance between image quality and imaging dose, a
commended that [7], in IGRT, minimizing CT dose must be within a CatPhan 500 phantom (Phantom Laboratory) which is 20 cm in dia-
context of relative cost versus benefit that will vary from patient to meter and includes a high spatial resolution module (CTP528) con-
patient. CT imaging parameters, including tube voltage (kV), tube sisting of a 1 through 21 line pair per centimeter test gauge as well as a
current (mA) and rotation time, slice thickness, and pitch, may be ad- low contrast module (CTP515) consisting of supra-slice and sub-slice
justed to optimize image quality versus radiation dose. IGRT with in- contrast targets was scanned with coaxial setups. CARE Dose4D was not
room CT imaging which offers diagnostic CT image quality has been turned on for image acquisition with this phantom because CARE
used in practice for online adaptive therapy [8]. However, extra time Dose4D has no current modulation effect over coaxial cylindrical ob-
spent on structure delineation for re-planning after image acquisition is jects. Scans were acquired with rotation time of 0.5 s, slice thickness of
still one of the major problems that make online adaptive therapy not 3 mm, pitch of 0.6, and the maximum available effective tube curren-
quite practical. Better image quality would help reducing the time t–time product with each of the following tube voltage settings: 70 kV
needed for the structure delineation. Similarly, for IGRT treatments, (415 mAs), 80 kV (541 mAs), 100 kV (541 mAs), 120 kV (514 mAs), and
any reduction in time spent for patient image registration verification 140 kV (451 mAs). The scans were repeated five times and then aver-
would help the ultimate treatment delivery accuracy. The purpose of aged to simulate high mAs scans which cannot be achieved due to
this study was to create a simple protocol for in-room CT imaging that machine limitation [15].
improves CT quality by optimizing CT acquisition and reconstruction After phantom data were acquired, the raw data were reconstructed
parameters including increasing CT dose if necessary for IGRT/treat- with different reconstruction algorithms. The reconstructed images of
ment verification as well as treatment planning (including adaptive re- the CTP 528 high spatial resolution module from all different tube
planning) in radiation therapy. voltage and effective tube current–time product settings were reviewed
by five experienced medical physicists. There is a tradeoff between
2. Materials and methods spatial resolution and noise for each reconstruction kernel. A smoother
kernel generates images with lower noise but with reduced spatial re-
2.1. General strategy solution. A sharper kernel generates images with higher spatial re-
solution, but increases the image noise. The selection of reconstruction
Images scanned with low dose protocols would introduce in- kernel should be based on specific clinical applications [16]. For our
accuracy in contouring which would lead to inadequate target coverage purpose, we reconstructed the CT images of the CTP 528 high spatial
or excess dose to organs at risk. However, if the image noise is small resolution module in the CatPhan 500 phantom with various re-
enough so that the contouring error from imaging is not the main construction kernels, and selected the first kernel (from smoother to
contributing factor, further increase of the image dose is not necessary. sharper) with reconstruction images where a 5 line pair (5/cm) in the
Therefore, for either treatment planning or IGRT, accurately deli- module could be clearly seen in all reconstructed images. To calculate
neating structures is our key indicator for the optimal protocol. CNR, two ROIs were drawn with the signal ROI inside the 15 mm dia-
Delineating accuracy is related to the image contrast and noise level. meter target in the 1% contrast level of the supra-slice in the CTP515
Contrast on the CT images was determined as the difference in CT module, and the background ROI outside but next to the target without
numbers between two materials [9]. The CNR [10,11] was defined as overlap over any other target in the CTP515 module. To determine the
the ratio of the contrast, which is the difference of mean CT numbers appropriate tube voltage, the CNR, the increased dose, as well as pe-
between the two regions of interest (ROI), and the average of standard netration was considered.
deviations of CT numbers in the two ROIs: In the meantime, in each phantom image, a circular ROI in the
uniform region was created and the CT numbers in the ROI were binned
|μS −μB | into a histogram. The distribution of the noise in uniform regions of the
CNR = 2 ,
σS + σB phantom was fit by a Gaussian. With the CatPhan 500 scan images at
different effective tube current–time product levels, the relationship
where μS and μB are the average CT numbers in the signal and back- between noise and the effective tube current–time product was de-
ground ROIs, and σS and σB are the standard deviations of CT numbers termined. The relationship was then used to simulate noise at certain
in the signal and background ROIs, respectively. To take radiation dose effective tube current–time product levels and the images with artifi-
into consideration, a dose-weighted CNR is defined as [12,13]: cially generated noise were compared against CatPhan 500 images that
CNR were taken later at the corresponding effective tube current–time pro-
CNRD = . duct.
√D

The maximum CNRD represents the minimum dose value for a given 2.3. Application of the protocol on patients
image quality level. Optimization is achieved with a maximum CNRD.
In this study, we used phantom data to find the optimal scanning The optimal parameters were obtained based on an analysis of the
parameters for our scan protocol. The protocol was then used on pa- phantom data. Image quality enhancement (IQE) protocols were then
tients. Patient scan data were analyzed to verify better contouring ac- created based on the optimized parameters. The IQE protocols suitable
curacy was achieved. for prostate and pancreas were used to acquire patient data at these two
All CT data were acquired with a CT scanner (Somatom Definition tumor sites for one daily IGRT scan per patient with the CT-on-Rails.
AS Open, Siemens) installed inside a LINAC room (i.e., CT-on-rails). The Higher tube voltage may be chosen in the IQE protocol for pancreas

7
G.-P. Chen et al. Physics and Imaging in Radiation Oncology 4 (2017) 6–11

Fig. 1. Variation of CNR (a) and CNRD (b) for the CatPhan 500 versus effective tube current–time product for different tube voltages. (c) The obtained relationship between noise and the
effective tube current–time product. (d) The CT number histograms (with mean CT number subtracted) from both simulation and phantom scan images at effective tube current–time
product of 100 mAs. The histograms were both normalized to the total counts.

patient to reduce the appearance of streaking artifacts caused by a deformable image registration using an in house tool based on a sys-
metal stent that usually exists in pancreas patients. Patient consent was tematic force Demons algorithm and a novel variable kernel smoothing
obtained. The image quality of the patient CTs acquired with IQE technique [17]. Contrast enhancement [18] and histogram matching
protocols was compared with that obtained with standard clinical [19] were used to further compensate for intensity differences between
protocols. An image pair, obtained from the standard clinical and IQE the CT sets acquired with different protocols. Registration accuracy was
protocols, was compared by surveying thirty radiation oncology staff then determined using the Dice similarity coefficient [20] to measure
including radiation oncologists, medical physicists, dosimetrists, and the overlap of planning and daily contours.
radiation therapists. The participants were asked to visually identify the As it is not practical to scan patients with a variety of effective tube
better image in terms of structure boundary and spatial resolution current–time product settings, we made an effort to simulate patient CT
without knowing the difference between the two. The comparison was scans with different tube current–time product values by adding arti-
also performed by evaluating the structure delineation consistency by ficially generated noise to the IQE protocol CT images. A Matlab routine
either (1) manual contouring of ten observers (two radiation oncolo- was written that generates noise using a random number generator. The
gists, two medical residents, two physicists, two dosimetrists and two relationship between noise and the effective tube current–time product
therapists), or with (2) an in-house deformable registration tool. was determined from the phantom data. The CT scans of the CatPhan
In method (1), contours for relevant structures (bladder, prostate 500 provided the noise amplitude, standard deviation, and noise power
and rectum for prostate patient, duodenum and pancreas head for spectrum as a function of effective tube current–time product. The
pancreas patient) were drawn on the two sets of CT images in two Matlab routine then extrapolated the observed noise properties into the
scenarios: (a) drawn ten times by one user (intra-observer) with one set desired imaging dose levels. Furthermore, a multi pixel smoothing was
of contours drawn a week after the previous set, and (b) drawn one time applied to the simulated noise in order to account for the observed
per participant by all ten observers (inter-observer). The variations noise texture. Once the noise model was set, it was applied to the
between the contours for a given organ were then compared by cal- question of optimal dose for the IQE protocol. The model was validated
culating the ratios of the standard deviation of volumes to the average with additional phantom scans. The improvement of CT quality can be
of the volumes of the contour. The average time spent on contouring measured by the ability to distinguish an edge between two organs in
one data set for all individual users was also compared for the image the body. The sharpness of the change in contrast as measured by the
data sets acquired with the standard clinical and IQE protocols. gradient is what makes an edge visible. The lower the contrast between
The purpose of method (2) was to find out if the IQE protocol can two organs the more difficult it is to delineate an edge, yet a low
improve image registration. The CT images acquired with the standard contrast edge can be distinguished if the gradient is sufficiently high.
clinical and IQE protocol during daily IGRT were registered to the Noise in the image degrades the gradient and hence the visibility of an
planning CT acquired with the standard clinical protocol based on edge. For our purposes, a useful test subject was the edge between the

8
G.-P. Chen et al. Physics and Imaging in Radiation Oncology 4 (2017) 6–11

Table 1 parameters were obtained. The CNRs increase with effective tube cur-
Imaging parameters for standard clinical and IQE protocols. rent–time product (Fig. 1a). Although not obvious, a weak trend with
the CNRD beginning to drop near 1000 mAs was seen (Fig. 1b). Based
Parameter Standard clinical IQE protocol
protocol on these, we determined that an effective tube current–time product of
Prostate Pancreas 1000 mAs was appropriate, because the dose weighted CNR, namely
CNRD, started to decrease or plateau. As a result of compromise, a tube
Tube voltage (kV) 120 100 120
voltage setting of 100 kV was optimal, but 120 kV was used for pan-
Effective tube current–time product 210 1000 1000
(mAs) creas patient as explained previously.
CareDose4D Off On On It was determined that the noise in uniform regions of the phantom
Rotation time (s) 0.5 1.0 1.0 was very well fit by a Gaussian which simplified the noise model. The
Pitch 0.6 0.6 0.6 relationship between noise and the effective tube current–time product
Slice thickness (mm) 3 3 3
Reconstruction algorithm B10f B31s B31s
determined is shown in Fig. 1c. We tested the model by rescanning the
Typical CTDIvol (mGy) 17 50 84 phantom. Fig. 1d shows the CT number histograms from both simula-
tion and phantom scan images at effective tube current–time product of
100 mAs. It was seen that the simulated noise spectrum was matched to
prostate and the bladder for the prostate patient. the noise spectrum shape as measured in the CatPhan 500 scans.

3. Results 3.2. Comparison of patient data with old and new protocols

3.1. Determination of optimal protocol and noise model based on phantom IQE protocols for patient scans were created based on phantom
data data. Sample IQE protocols suitable for prostate and pancreas scans
were compared with the standard clinical protocols in Table 1. The
The reconstruction kernel B31s was found to be the first sharper main differences between the standard clinical and IQE protocols were
kernel with which the 5 line pair was all clearly seen in the re- that the effective tube current–time product was increased to
constructed images, and it was used in all phantom image reconstruc- 1000 mAs, CareDose4D was turned on, the rotation time was increased
tion. The CNRs for the target in CatPhan images with different imaging from 0.5 s to 1.0 s, and the reconstruction algorithm was changed from

Fig. 2. Sample images scanned with standard clinical (a) and IQE (b) protocols for a prostate cancer patient. Contour variation (ratios of standard deviation over average of structure
volumes) for the same patient: intra-observer (c) and inter-observer (d).

9
G.-P. Chen et al. Physics and Imaging in Radiation Oncology 4 (2017) 6–11

Fig. 3. A sample image (b) with noise level corresponding to 500 mAs, which was compared with the CTs acquired with the standard clinical (a, 210 mAs) and IQE (c, 1000 mAs)
protocols. The line in (b) represents a strip of voxels which illuminate the gradient between prostate and bladder as described in the text. (d) The CT number crossing the boundary in
images acquired with the IQE protocol (High), the standard clinical protocol (Low) and the simulated image generated from adding noise to the IQE protocol (Adjusted).

B10f to B31s based on the phantom data and was verified with patient images from IQE protocol with planning CT images were found to be
image data. Increasing rotation time allowed for larger effective tube slightly larger than those for the registration of CT images from the
current–time product to be used so that a patient scan of 1000 effective standard clinical protocol with the same planning CT, e.g., by 3.3
mAs could be performed in one scan. percentage point (pp) for bladder, 1.0 pp for prostate, and 1.1 pp for
Visual inspection of the CT images obtained with IQE protocol de- rectum, indicating again that the IQE protocol improved CT quality for
monstrated improved visibility of the prostate gland boundary. An IGRT. The above results for improved registration with the IQE protocol
image pair, obtained from the standard clinical and IQE protocols, was were further examined by assessing the enhanced edge delineation due
shown in Fig. 2. A majority of the survey participants (twenty seven out to noise reduction. Fig. 3 shows a sample image (b) with artificial noise
of thirty) indicated that the image produced with the IQE protocol was level corresponding to 500 mAs, which was compared with the CT
better. The averaged reduction of time for contouring bladder, prostate images acquired with the standard clinical (a, 210 mAs) and IQE (c,
and rectum on images acquired with the IQE protocol compared with 1000 mAs) protocols. As a measure of the image quality in these three
that on images acquired with the standard clinical protocol was ∼ 10%. images, we extracted the CT numbers in a strip of voxels that cross the
For intra-observer variation, the IQE protocol yielded smaller variations boundary between the prostate and bladder (the vertical line as shown
for all three structures for the prostate cancer case (Fig. 2c). For the in Fig. 3b). Plots of the voxel location versus CT number are presented
inter-observer variations, the ratios for both prostate and bladder were in Fig. 3d for the three CT sets. It was seen that as the effective dose
smaller for the IQE protocol; however for the rectum the ratio was increases, the noise is reduced and it has the tendency that the CT
slightly higher for the IQE protocol in comparison to the standard number has higher gradient at the location of the edge. Thus, the higher
clinical protocol (Fig. 2d). This exception may be partially explained by effective dose CT achieves a meaningful enhancement in the delinea-
the substantial differences in the shape and volume of the rectum be- tion of structures.
tween the two days. The Dice coefficients for the registration of CT The same comparisons were also made for a pancreatic cancer

10
G.-P. Chen et al. Physics and Imaging in Radiation Oncology 4 (2017) 6–11

patient. Visual inspection also indicated that the quality of the image (∼7 cGy extra CT dose), for online adaptive replanning together with
scanned with the IQE protocol was improved over the image scanned IGRT use (∼7 cGy extra each time for 2–3 times in the course of
with the standard clinical protocol. The dice coefficients for pancreas treatment depending on patient anatomy change), for weekly IGRT
were calculated for 5 slices. The average improvement in the registra- (∼35 cGy extra), or for hypo-fractionated treatment such as SBRT
tion was 8.5 pp from the CT images with the standard clinical protocol treatments delivered in three to five fractions (∼35 cGy extra), the dose
to those with the IQE protocol. increase as compared to the prescription dose would be acceptable.
CT protocols can be optimized for radiation therapy planning and
4. Discussion delivery by considering specific situations including tissue types (elec-
tron densities) and tumor sites (e.g., body thickness). When all factors
In this study, we created IQE protocols based on phantom mea- are considered, tube voltage of 100 or 120 kV and effective tube cur-
surements and applied them with patient scans to improve image rent–time product of 1000 mAs were found to be the appropriate
quality. Whilst most other studies focused only on either treatment choices for scanning abdomen and pelvis. The CT image quality with
planning contours or IGRT, we considered treatment planning contours IQE protocols was improved comparing to those using the standard
for both initial planning and online adaptive replanning purposes, as clinic protocols. The imaging dose with the IQE protocol would be
well as IGRT. Therefore, time for structure delineation and image re- approximately three to five times of that with the standard clinic pro-
gistration was also one of our concerns, and we did not use treatment tocols, which may be justified for RT, particularly for RT planning,
plan quality as a metric for the image quality but used the CNRD re- online adaptive RT and SBRT.
sponse with effective tube current–time product instead. The study by
Li et al proposed and implemented a general strategy for automatic CT Acknowledgement
simulation technique selection for radiation therapy by using the IQI
which was defined to characterize the simulation performance on This study was supported in part by Medical College of Wisconsin
structure delineation and used to benchmark the contouring accuracy Cancer Center Fotsch Foundation and by Siemens Medical.
and treatment plan quality [5]. Our study focused more on the treat-
ment delivery accuracy and was not intended for individual patient. References
This is because for treatment delivery, setup time is often limited;
protocol adjustment per patient is subject to human error. Unlike [1] Brenner D. Induced cancers after prostate-cancer radiotherapy: no cause for con-
Westerly et al.’s study, we used in-room CT scanner which provides cern? Int J Radiat Oncol Biol Phys 2006;65(3):637–9.
[2] Kalra MK, Maher MM, Toth TL, et al. Strategies for CT radiation dose optimization.
diagnostic level image quality and could benefits treatment planning Radiology 2004;230(3):619–28.
and both online and offline adaptive therapy also. [3] Hevezi JM, Mahesh M. Optimizing CT dose and image quality for radiotherapy
Optimization of imaging is one of the three major areas that could patients. JACR 2012;9(2):152.
[4] Njeh CF, Dong L, Orton CG. IGRT has limited clinical value due to lack of accurate
contribute to improving the accuracy of structure delineation in addi- tumor delineation. Med Phys 2008;33:136–40.
tion to implementation of standardized delineation protocols and [5] Li H, Yu L, Anastasio MA, et al. Automatic CT simulation optimization for radiation
guidelines, as well as specialized training [21]. While the latter two are therapy: a general strategy. Med Phys 2014;41(3):031913.
[6] Westerly DC, Schefter TE, Kavanagh BD, et al. High-dose MVCT image guidance for
not in the scope of our discussion, image quality improvement can be stereotactic body radiation therapy. Med Phys 2012;39(8):4812–9.
achieved with new imaging techniques such as MRI, PET-CT, and [7] Murphy MJ, Batler J, Batler S, et al. The management of imaging dose during
functional MRI that improve visibility of the target. Potential ad- image-guided radiotherapy: report of the AAPM Task Group 75. Med Phys
2007;34(10):4041–63.
vantages of functional imaging modalities are reduction of inter-ob-
[8] Ahunbay E, Peng C, Chen GP, et al. An on-line replanning scheme for interfractional
server variability, identification of tumor extensions missed by CT and/ variations. Med Phys 2008;35(8):3607–15.
or MRI and possibly identification of GTV subvolumes requiring higher [9] Boone JM, Geraghty EM, Seibert JA, Wootton-Gorges SL. Dose reduction in pedia-
radiation dose. However, most of the current IGRT techniques are still tric CT: a rational approach. Radiology 2003;228(2):352–60.
[10] Yan H, Cervino L, Jia X, Jiang S. A comprehensive study on the relationship be-
CT based. Our study showed evidence that the inter-observer variability tween the image quality and imaging dose in low-dose cone beam CT. Phys Med
could be reduced with improved image quality by using the IQE pro- Biol 2012;57(7):2063–79.
tocols. [11] Xu J, Reh DD, Carey JP, Mahesh M, Siewerdsen JH. Technical assessment of a cone-
beam CT scanner for otolaryngology imaging: image quality, dose, and technique
The use of IQE protocols on patient scans showed some improve- protocols. Med Phys 2012;39(8):4932–42.
ment in image quality over a standard clinical protocol. However, the [12] Kalender WA, Deak P, Kellermeier M, van Straten M, Vollmar SV. Application- and
imaging dose, based on CTDIvol, to patients with an IQE protocol was patient size-dependent optimization of x-ray spectra for CT. Med Phys
2009;36(3):993–1007.
about three to five times of that with the standard clinical protocol, [13] Lee SW, Kim HJ, Kim DH, Lee CL. Evaluation of dose reduction and image quality in
namely, about 50–84 mGy per scan. Increasing imaging dose to improve pediatric multi-detector CT. J Korean Phys Soc 2010;57(1):183–91.
image quality must be weighed against the risks associated with addi- [14] Flohr T. CARE Dose 4D. White Paper (Siemens Healthcare) 2011; A9115–111236-
C1-4A00.
tional exposure to the patient (especially for pediatric patients). As [15] Hulme KW, Rong J, Chasen B, et al. A CT acquisition technique to generate images
Westerly et al. pointed out [6], assessing the risks posed by imaging is at various dose levels for prospective dose reduction studies. AJR Am J Roentgenol
difficult because the risks that come with the additional radiation ex- 2011;196(2):W144–51.
[16] Computed Body Tomography with MRI Correlation, 4th Edition, Volume 1, edited
posure (e.g. to bladder and rectum for prostate patients) are offset by
by Joseph K. T. Lee, Lippincott Williams & Wilkins, 2006. 1821p.
the potential benefits of early detection (in the case of diagnostic x-ray [17] Hart V, Chen GP, Li XA. The effect of CT image quality on deformable image re-
imaging) and more precise targeting (in the case of treatment planning gistration in radiotherapy. RSNA 2013.
and IGRT). The three steps to manage imaging dose recommended by [18] Ritika Kaur S. Contrast enhancement techniques for images-a visual analysis. Int J
Comp Appl 2013;64(17):20–5.
AAPM TG75 [7] can be followed: (1) assessment, (2) reduction, and (3) [19] Coltuc D, Bolon P, Chassery JM. Exact histogram specification. IEEE Trans Image
optimization. If the IQE protocol is used for all fractions of a prostate Process 2006;15(5):1143–52.
patient with conventional fractionation, the CT dose would be a little [20] Allozi R, Li XA, White J, et al. Tools for consensus analysis of experts’ contours for
radiotherapy structure definitions. Radiother Oncol 2010;97(3):572–8.
bit over 3 Gy which would be significant as compared to the prescrip- [21] Segedin B, Petric P. Uncertainties in target volume delineation in radiotherapy-are
tion dose, and therefore it might not be suitable for daily IGRT for a they relevant and what can we do about them? Radiother Oncol
standard treatment plan. However, if IQE protocol is used instead of 2016;50(3):254–62.
standard clinical protocol for treatment planning CT acquisition

11