survey of computed tomography scanners in taiwan: dose descriptors, dose guidance levels, and...

Survey of computed tomography scanners in Taiwan: Dose descriptors,dose guidance levels, and effective doses

H. Y. TsaiDepartment of Medical Imaging and Radiological Sciences, Chung Shan Medical University,Taichung 402, Taiwan and Department of Radiology, Chung Shan Medical University Hospital,Taichung 402, Taiwan

C. J. Tunga�

Department of Biomedical Engineering & Environmental Sciences, National Tsing Hua University,Hsinchu 300, Taiwan

C. C. YuDepartment of Radiological Technology, Yuanpei University, Hsinchu 300, Taiwan

Y. S. TyanDepartment of Radiology, Chung Shan Medical University Hospital, Taichung 402, Taiwan

�Received 30 August 2006; revised 19 January 2007; accepted for publication 7 February 2007;published 14 March 2007�

The IAEA and the ICRP recommended dose guidance levels for the most frequent computedtomography �CT� examinations to promote strategies for the optimization of radiation dose to CTpatients. A national survey, including on-site measurements and questionnaires, was conducted inTaiwan in order to establish dose guidance levels and evaluate effective doses for CT. The beamquality and output and the phantom doses were measured for nine representative CT scanners.Questionnaire forms were completed by respondents from facilities of 146 CT scanners out of 285total scanners. Information on patient, procedure, scanner, and technique for the head and bodyexaminations was provided. The weighted computed tomography dose index �CTDIw�, the doselength product �DLP�, organ doses and effective dose were calculated using measured data, ques-tionnaire information and Monte Carlo simulation results. A cost-effective analysis was applied toderive the dose guidance levels on CTDIw and DLP for several CT examinations. The meaneffective dose±standard deviation distributes from 1.6±0.9 mSv for the routine head examinationto 13±11 mSv for the examination of liver, spleen, and pancreas. The surveyed results and the doseguidance levels were provided to the national authorities to develop quality control standards andprotocols for CT examinations. © 2007 American Association of Physicists in Medicine.�DOI: 10.1118/1.2712412�

I. INTRODUCTION

The largest man-made source of ionizing radiation exposureto the population is from diagnostic medical x rays.1 Com-puted tomography �CT� constitutes the largest contribution tothe radiation exposure of the population from diagnosticmedical sources.2 Therefore, efforts towards dose reductionin CT have been recommended by the international organi-zations such as the International Commission on Radiologi-cal Protection �ICRP�. The ICRP,3–5 the International AtomicEnergy Agency �IAEA�,6 and the European Commission�EC�7–9 have all recommended the setup and the implemen-tation of CT dose guidance levels for the most frequent ex-aminations to promote strategies for the optimization of CTdoses.7 These dose guidance levels should be derived usingdata from a wide scale survey.

Effective dose data or any dose descriptors per CT exami-nation are available but rare. Their values depend on thescanner, examination, protocol, and conditions of the opera-tion. The increased speed and the use large volume scanswith thinner slices tend to increase the radiation exposurefrom a particular CT examination. Advanced CT techniques
such as spiral CT, electron beam CT, or multi-slice CT add to
1234 Med. Phys. 34 „4…, April 2007 0094-2405/2007/34„4…/

the complexity of the dose evaluations even for similar CTexamination protocols �scan length, slice thickness, etc.�.The Imaging Performance Assessment of CT Scanners �Im-PACT� group has provided data for the commonly used mod-els of CT scanners. For other scanners, the ImPACT devel-oped a method10 by matching the dosimetric characteristicsof the scanner models to those scanner models covered bythe National Radiological Protection Board �NRPB� datasets.11,12 In this way, a new scanner model can be matched toan existing Monte Carlo �MC� data set that describes its dosedistribution. Measured computed tomography dose index�CTDI� and NRPB MC data sets for the total normalizedorgan dose for the scanned volume were then used to evalu-ate the CT organ doses and the effective dose.

A national survey was carried out in Taiwan to estimateradiation doses to CT patients and to establish dose guidancelevels. This survey consisted of two parts, i.e., on-site mea-surements and questionnaires. The beam quality and outputand the phantom doses were measured for nine representa-tive CT scanners. Results of measurements were comparedto corresponding data provided by the ImPACT.13 Question-
naire forms were sent to all �285 total� scanners in Taiwan
12341234/10/$23.00 © 2007 Am. Assoc. Phys. Med.

http://dx.doi.org/10.1118/1.2712412

http://dx.doi.org/10.1118/1.2712412

http://dx.doi.org/10.1118/1.2712412

1235 Tsai et al.: Survey of CT scanners in Taiwan 1235

and completed by respondents from facilities of 146 scan-ners. Information on patient, procedure, scanner, and tech-nique for the head and body examinations and on CT qualitycontrol program was provided. The weighted CTDI, thedose-length product �DLP�, the organ doses and the effectivedose were then calculated using measured data, question-naire information, and MC simulation results.11,12 A cost-effective analysis14 was applied to derive the national doseguidance levels of various dose descriptors.

II. METHODS AND MATERIALS

A. Dose descriptors measurements

Measurements of dose descriptors were performed in nineCT scanners according to the European Guidelines7 and themeasurement protocol established by the ImPACT and St.George’s Healthcare.15 The manufactures of these nine scan-ners which are the most common in Taiwan, are GeneralElectric for five scanners, Siemens for two, Picker for one,and Hitachi for one.

A calibrated �National Radiation Standard Laboratory, In-stitute of Nuclear Energy Research, Taiwan� pencil-type ion-ization chamber �Model 660-6; Victoreen, Cleveland, OH�with 10 cm sensitive length, and a digital exposure meter�Model 660; Victoreen, Cleveland, OH� were used to mea-sure the CTDI values. The CTDI in mGy �absorbed dose toair� is given by

CTDI = Xr · Cc · f · L/T , �1�

where Xr is the electrometer reading in R; Cc, the calibrationcorrection factor; f , the conversion factor of 8.76 mGy/R; L,the chamber sensitive length of 10 cm; and T, the nominalslice thickness or beam collimation.16 The standard head andbody CT dosimetry phantoms �Model 660-7: head, Model660-8: body; Victoreen, Cleveland, OH� are made of cylin-drical polymethyl methacrylate �PMMA� with 16 and 32 cmin diameter and 15 cm in length.

The CTDIs were measured along the central axis of thegantry, free in air �CTDIair�, and along the central �CTDIc�and peripheral �CTDIp� axes of each 16 and 32 cm dosimetryphantom. The measurements were performed under severalclinical technical factors in head or body examinations. Theweighted CTDI �CTDIw� as defined by the InternationalElectrotechnical Commission.7 was then calculated from

CTDIw = 13CTDIc + 2

3CTDIp, �2�

where CTDIp is the average of all four peripheral CTDIs�top, left, right, and bottom with respect to the couch�. Thenormalized CTDIw �nCTDIw� in mGy/mAs or mGy/100 mAsis the CTDIw normalized to tube current-time product perrotation.

After measuring the CTDIair with several collimations for
clinical uses, the collimation factor �fcol� is then given by
Medical Physics, Vol. 34, No. 4, April 2007

fcol =CTDIair,col

CTDIair,10 mm, �3�

where CTDIair,col is the CTDIair at available collimation; andCTDIair,10 mm, the CTDIair at 10 mm collimation. The colli-mation factor could be applied to the measurements madeonly at the 10 mm collimation, as well as CTDIs other thanthe 10 mm collimation.

B. Questionnaire survey

The purpose of the questionnaire survey was to determinethe extent of CT activities in Taiwan and to estimate patientdoses caused by CT practice. The content of the question-naire was in accordance to the European Guidelines7 andNRPB reports.17,18 Before beginning a nationwide survey,questionnaire trials were done by several hospitals, whosesuggestions were included in the final form of the question-naire. Each questionnaire included the following informa-tion: �a� hospital, scanner manufacturer, scanner model, andinstalled year; �b� the utilization rate, the maintenance, andthe quality assurance; �c� the sketch of a mathematicalphantom10 with a grid indicating the positions of 5 mm thickslabs for specification of the scan range, which might becomposed of different scan sequences; �d� for each scan se-quence, the number of slices, slice thickness, couch incre-ment, pitch, tube voltage, tube current, and time per rotation;and �e� for each scan sequence, a condition of whether con-trast agents were used or not.

C. Determining the guidance levels

The EUR 162627 specified the CTDIw and the DLP, de-fined by the International Electrotechnical Commission,19 asreference dose descriptors for CT practice.7 The CTDIw andthe DLP were calculated using measured data and question-naire information. The CTDIw obtained from Eq. �2� repre-sents the average dose to a phantom for a single axial scan.The DLP in mGy cm, characterizing the exposure for a com-plete CT examination, was calculated from

DLP = �i

�nCTDIw · T · N · C�i �4�

or

DLP = �i

�nCTDIw · T · A · t�i, �5�

where i represents each serial or helical scan sequence form-ing part of one examination; nCTDIw, the CTDIw normalizedto the tube current-time product; T, the slice thickness; N, thenumber of slices; C, the tube current-time product �mAs�; A,the tube current �mA�; and t, the total acquisition time �s� forthe sequence.7

The Institute of Physical Sciences in Medicine �IPSM�has recommended the third quartile values of dose descrip-tors as the dose guidance levels.20 However, the guidancelevel derived by a cost-effective approach conforms more tothe optimization considerations.14 A cost-effective approach
is to plot the curve of the accumulating percentage of scan-


ners with reference doses, where the slope of the curve rep-resents the relative number of scanners per unit interval ofreference dose. Thus a steeper slope corresponds to a moreeffective guidance level. In this study, the intersection of twochanging slopes is selected as the guidance level.

D. Dose assessments

The organ doses were estimated by using questionnaireinformation, nCTDIair values from measurements, ImPACTdata, and the MC data sets in NRPB-SR250.11,12 through theImPACT calculator.13 The NRPB MC data sets included 23operating conditions of 5 manufacturers. Each MC data setcontains doses of 27 organs and regions, which are obtainedfrom the simulated individual irradiation of 208 contiguous5 mm transverse slices with perfect collimation, covering themathematical phantom. Each organ dose �HT� from a CTscan series is derived from

HT = nCTDIair · C · fcol ·T

I· ��en

��

air

tissue

· � vDi, �6�

where I is the table movement per rotation; ��en/��airtissue, the

ratio of mass energy absorption coefficients for the tissue toair; Di, the organ dose by the irradiation of i scan slice; �vDi,the total organ dose summed over the scan volume, v.18 Theeffective dose �E� defined by the ICRP21 is then given by

E = �T

wTHT, �7�

where wT is the weighting factor for tissue T.The matching method developed by the ImPACT10,22 is to

appropriately select a MC data set to represent a newer scan-ner, which is not included in NRPB-SR250, by ImPACT fac-tors �ImF�. The ImF is derived from

ImF = �� CTDIc

CTDIair� + �� CTDIp

CTDIair� + � , �8�

where � is 0.4738 �head�, 3.5842 �body�; �, 0.8045 �head�,0.6238 �body�; and �, 0.0752 �head�, −0.0902 �body�.

E. Data processing and statistical analysis

More than 900 calculations of organ doses and effectivedoses were performed for several different CT examinations.The results from the ImPACT calculator, including total ef-fective dose, gonad dose, red bone marrow �RBM� dose,CTDIw and DLP, were recorded into the Excel database. Lin-ear regression between the DLP and the effective dose wasperformed using the SigmaPlot version 9 to derive theregion-specific normalized effective dose �EDPL�.7 The rela-tionship is given by

E = EDLP � DLP. �9�

The EDPL is the slope of the straight-line regression and wascompared to those reported in the EUR 162627 and Italian

23
survey.

III. RESULTS AND DISCUSSION

Questionnaires were sent to all 285 CT facilities in Tai-wan and returned by 146 respondents. Questionnaires werecompleted according to the examinations of a representativeadult patient. Table I summarizes the distribution of manu-facturers and models, the model reference name referred inthe present work, the ImPACT code, and the number of com-pleted questionnaires for each model. There were 20 scannermodels from six manufacturers as General Electric �40%�,Siemens �16%�, Toshiba �21%�, Shimadzu �2%�, Picker�5%�, and Hitachi �17%� in this national survey of Taiwan.The present measurements involved nine representative

FIG. 1. The measured �solid triangles� CTDIair and CTDIw per 100 mAs at10-mm beam collimation width for different scanners with available kVpsettings. Corresponding data �open circles� reported by the ImPACT groupare included for comparison. �a� Normalized CTDIair; �b� normalized CTDIw

in the head phantom; �c� normalized CTDIw in the body phantom.

scanners belonging to four manufacturers. Table II shows the


TABLE I. Summary of 146 CT scanners in this study including the manufacturer and model name, the modelreference name referred in the present work, the ImPACT code, and the number of completed questionnaires foreach model.

Manufacturer Model

Modelreference

nameImPACT

code

Numbercompleted

questionnaires

HiLight, HiSpeed, CT/i�without SmartBeam�

G1 GE.c 8

HiSpeed CT/i withSmartBeam

G2 GE.d 9

General Max G3 GE.e 9Electric Pace, Sytec GE.f 12

Prospeed G4 GE.g 7FX/i, LX/I G5 GE.h 12

QX/i, LightSpeed,LightSpeed Plus

GE.i 1

Somatom Plus 4 Series SI.d 5

Somatom AR-C, AR.SP,AR-T

SI.e 7

Siemens Plus, DXP, Plus-S SI.g 1Hi Q SI.h 1

Volume Zoom, Access S1 SI.j 4Emotion Duo S2 SI.k 5

Toshiba TCT 600 TO.a 19

Xvision / EX TO.d 4Xpress HS1 TO.e 1

Auklet TO.i 6Shimadzu SCT SH.a 3

Picker PQ series P1 PI.b 7

HITACHI CT-W series H1 – 25

Total 146

TABLE II. Measured CTDIair, CTDIc, and CTDIp and calculated CTDIw for head and body scans with 10 mm collimation and available tube voltages.

Scanner kVpHead Body Head Body

nCTDIair nCTDIc nCTDIp nCTDIair nCTDIc nCTDIp nCTDIw

G1 120 18.9 12.0 12.6 18.9 3.2 5.8 12.4 4.9G1 140 25.9 17.2 17.5 – – – 17.4 –G2 120 19.9 12.3 12.9 19.8 3.4 6.2 12.7 5.3G2 140 27.3 17.6 18.0 27.2 5.2 9.2 17.9 7.8G3 120 34.3 18.3 19.9 33.8 4.5 9.9 19.4 8.1G4 120 36.6 19.0 20.7 36.6 5.6 11.2 20.2 9.4G5 120 35.3 20.2 21.6 35.3 5.3 11.0 21.1 9.1G5 140 45.4 28.0 29.4 45.4 7.8 14.5 28.9 12.3S1 120 23.9 16.8 19.8 17.3 4.4 9.6 18.8 7.8S1 140 32.9 23.7 27.5 25.0 7.0 14.3 26.2 11.8S2 110 20.9 13.6 14.7 21.0 3.8 7.8 14.4 6.4S2 130 29.4 20.0 21.3 29.5 6.1 11.7 20.9 9.8P1 130 34.6 18.0 16.8 34.6 6.1 15.4 17.2 12.3P1 140 38.8 20.3 18.7 38.8 7.4 17.8 19.2 14.3H1 120 17.4 11.5 12.4 17.3 3.8 7.7 12.1 6.4



results of measurements on nCTDIair, nCTDIc, and nCTDIp at10 mm collimation, and the calculated nCTDIw for head andbody scans. The values of nCTDIair are approximately thesame for head and body scans, except for scanner S1 withsmaller values in body scan due to different bow-tie filters.The values of nCTDIc and nCTDIp are smaller in the bodyphantom than in the head phantom due to more attenuationof x rays.

Figure 1�a� plots the measured values of nCTDIair �solidtriangles� at 10-mm beam collimation for different scannerswith available kVp settings. The values of nCTDIair fluctu-ated within 0.8 % for clinical tube current range and within0.5% for clinical exposure time range. The nCTDIair is ap-proximately proportional to the square of the tube voltage forscanner G1, G2, S1, and S2. But the relationship is slightlydifferent for scanners G5 and P1. The values of nCTDIw�solid triangles� in head and body phantoms were measuredat the same conditions as nCTDIair, and plotted in Figs. 1�b�and 1�c�, respectively. ImPACT provided data13 on all scan-ner models included in this study, except Hitachi. Our mea-surements of nCTDI values were compared with correspond-ing ImPACT data �open circles� in Fig. 1. The relativedifference between measured and ImPACT data, i.e., �mea-sured data–ImPACT data� / ImPACT data, is within ±10%.

TABLE IV. Nominal technical factors and dose descriptors for CT examinati

Examination

Nominal technical facto

Tube voltage�kVp�

Tube current-timeproduct �mAs�

Mean Range Mean Range Me

Routine Head 121 100–140 343 90–1500Routine Chest 122 120–140 268 40–1350Routine Abdomen 121 120–140 292 60–1575Routine Pelvis 121 120–140 295 60–1350LSPb 121 120–130 316 60–1500Kidney 121 120–130 322 60–1350

aData are mean values ± standard deviations.bLSP: liver, spleen, and pancreas.c

TABLE III. Collimation factors normalized to 10-mm

Collimation�mm� G1 G2 G3

201610 1.00 1.00 1.0087 1.00 1.005 1.00 1.02 1.0343 1.05 1.052 1.011.51 1.12 1.13

Data are the results from complete examinations with one or more than one seq


Note that the ImPACT data are average values from a survey,but there are variations from one scanner to another. Al-though the variation exists, the relative difference is accept-able. For those scanners not included in our measurements,the ImPACT data sets were used to estimatenCTDIw.

The collimation factor �fcol� in Table III is used to con-sider the output change with x-ray collimated beam width.The collimation factor distributes from 0.94 to 1.05 for thecollimation between 1.5 to 10 mm, except scanner S1 withcollimation factor 1.20 at 4 mm collimation and 1.30 at2 mm collimation, and scanner H1 with collimation factor0.91 at 2 mm collimation. As the collimation reduces to1 mm for high resolution CT �HRCT�, the collimation factorbecomes relatively high except scanner P1 and ranges be-tween 1.12 and 2.86. Scanner S1, with a collimation factor of0.88 for 16 mm and 0.92 for 20 mm, is a multislice CT scan-ner. The results indicate that most scanners have well-matched collimation, i.e., with collimation factor close to 1,for larger beam width, but not good collimation for the nar-rowest beam width. Furthermore, the dose evaluation with-out considering the collimation factor could underestimatemore than two times for the narrowest beam, but only devi-ate ±6% for 3- to 10-mm collimation. All collimation factors

tudied in the present work.

Dose descriptors

n rangecm�

Scansequencea

CTDIw

�mGy�DLP

�mGy cm�

Range Mean±SD Mean Range Mean Range

7.5–21 55 18–181 665 183–21736–45 20 3–86 455 50–2157

10–35 22 4–101 453 58–18987.5–40 22 4–86 410 67–19847.5–36 2.3±0.8 23 4–96 758c 97–2876c

3–25 2.0±0.6 24 4–86 617c 47–2157c

mation.

Scanner

5 S1 S2 P1 H1

0.920.88

.00 1.00 1.00 1.00 1.001.02 0.95

1.04 1.03 0.96 0.971.20 0.96

1.01 0.95 0.95.01 1.30 1.02 0.94 0.91

0.942.08 1.27 0.95 2.86

ons s

rs

Sca�

an

12.222.120.418.717.213.3

colli

G

1

1

uence according to the protocol in each hospital.


agree well with ImPACT data. Based on the above measure-ments and evaluations, the ImPACT database was, therefore,used to estimate the patient dose if scanners were includedthere. Otherwise, measured data on CTDIair, CTDIc, andCTDIp for the scanner type were used instead. To estimatepatient dose using ImPACT packages, additional informationon scanning parameters such as mAs, slice width, collima-tion, number of slices, scan start, and end position, andcouch increment was required. Table IV shows the summaryof technical factors and dose descriptors for several routineCT examinations collected from the present survey. Themean values, ranges, and standard deviations of these factorsand descriptors are given in this table.

Figure 2 �top panel� illustrates the distribution of CTDIw

for the routine head examination. It reveals that this distribu-tion is roughly a log-normal distribution, with few scannershaving much greater CTDIw than others. The first quartile,the median, and the third quartile values of this distributionare 35, 49, and 62 mGy, respectively. Although the thirdquartile value was recommended2 as the dose guidance levelby the NRPB, one may derive this level based on the opti-mization considerations.14 Figure 2 �bottom panel� plots theaccumulating percentage of scanners �ordinate� with CTDIw

below a certain value �abscissa�. More scanners meet thedose constraint by selecting a larger CTDIw as the guidancelevel. Decreasing the guidance level will increase the effortto bring more scanners within the dose constraint. Since theslope of the curve in Fig. 2 represents the relative number ofscanners per unit interval of CTDIw, the steeper the slopeindicates the more effective the guidance level. Therefore, acost-effective approach to determine the guidance level is toselect CTDIw at the intersection between two lines �dashedlines� of rapidly changing slopes. This point, correspondingto 72 mGy �see the right arrow tick labeled “a” on abscissa�,covers 85% of scanners and is greater than the guidance levelof 60 mGy �see the left arrow tick labeled “c”� recom-mended by the EC. Here we suggest that a less restrictiveguidance level, i.e., 72 mGy, be implemented in Taiwan atthe present stage. Similarly, Fig. 3 �top panel� plots the dis-tribution of surveyed data on DLP for the routine abdomen

TABLE V. Comparison of guidance levels on CTDrecommended by the EC.

Guidance level on CT�mGy�

ExaminationCost-effective

approachaThird

quartil

Routine head 72 �85%� 62Routine chest 24 �83%� 21Routine abdomen 31 �90%� 23Routine pelvis 28 �83%� 23LSPb 24 �81%� 23Kidneys 24 �80%� 23

aData in parentheses are percentages of scanners witbLSP: liver, spleen, and pancreas.cData are the results from complete examinations with
in each hospital.

examination. Again, an approximate log-normal distributioncan be seen. The dose guidance level recommended by theEC is 780 mGy cm. Figure 3 �bottom panel� plots the accu-mulating percentage of scanners �ordinate� with DLP belowa certain value �abscissa�. Here the cost-effective determina-tion of the guidance level is to select the DLP at either pointA or B. If point B is selected, the guidance level corresponds

d DLP determined in the present study and those

Guidance level on DLP�mGy cm�

ECCost-effective

approachaThird

quartile EC

60 850 �83%� 763 105030 580 �80%� 535 65035 680 �87%� 500 78035 520 �83%� 459 57035 800 �73%�c 856c 900– 890 �83%�c 658c –

es below the determined guidance levels.

or more than one sequence according to the protocol

FIG. 2. �Top panel� A plot of the distribution of CTDIw for the head exami-nation, revealing roughly a log-normal distribution. �Bottom panel� A plot ofthe accumulating percentage of scanners with CTDIw below a certain value�abscissa�, where the slope of the curve represents the relative number ofscanners per unit interval of CTDIw. Determining the guidance level by acost-effective approach is to select CTDIw at the intersection of two lines�dashed lines� of rapidly changing slopes. The CTDIw of this intersection iscorresponding to 72 mGy �see the right arrow tick labeled “a” on abscissa�,which covers 85% of scanners, and greater than the guidance level of60 mGy �the left arrow tick labeled “c”� recommended by the EC. The thirdquartile of CTDIw in this survey is 62 mGy �the middle arrow tick labeled“b”�.

Iw an

DIw

e

h dos

one


to DLP close to the third quartile value. Alternatively, pointA may be selected which covers 87% of scanners with DLPbelow the guidance level. Still we suggest that the less re-strictive value, i.e., 680 mGy cm, be implemented at thepresent stage. Table V is a comparison of the dose guidancelevels on CTDIw and DLP determined in this study and rec-ommended by the EC. Although these examinations had dif-ferent contrast techniques applied �without contrast, withcontrast, or both�, the guidance levels in Table V representvalues for one sequence or phase of each routine examina-

FIG. 4. The derived ImPACT factors �solid triangles� of head and body s

FIG. 3. �Top panel� a plot of the distribution of DLP for the liver, spleen, andpancreas �LSP� examination, revealing roughly a log-normal distribution.�Bottom panel� a plot of the accumulating percentage of scanners with DLPbelow a certain value �abscissa�, where the slope of the curve represents therelative number of scanners per unit interval of DLP. Determining the guid-ance level by a cost-effective approach is to select DLP at the intersection,either A or B, between two lines �dashed lines� of rapidly changing slopes.The DLPs of intersection A and B are corresponding to 500 �see the leftarrow tick on abscissa� and 680 mGy cm �the middle arrow tick�, whichcover 75% �third quartile� and 87% of scanners as compared to the guidancelevel of 780 mGy cm �the right arrow tick� recommended by the EC.

ImPACT group are included for comparison. �a� ImPACT factors of head scanni


tion. For liver-spleen-pancreas and kidney examinations, theguidance levels are determined by one complete examinationwith more than one sequence.

Table VI shows the ImPACT factor and the matching MCdata set for head and body scans. A higher ImF means aharder x ray beam.24 Figure 4 shows a comparison of theImF from present measurements and ImPACT database. TheImF values seem to agree with each other quite well. But theMC data set chosen from our measurements is not all thesame with that reported by ImPACT in Table VI. Further-more, the selection of MC data sets affects the calculationsof organ doses and the effective dose. For example, the dif-ference between the effective doses calculated from MC dataset 8 and 10 for scanner G3 is about 18% for routine headexaminations with the same setting of technical factors. Also,the difference of effective doses, calculated from MC data

for different scanners. Corresponding data �open circles� reported by the

TABLE VI. The calculated ImPACT factor and the matching MC data set foreach scanner with available tube voltages.

ImFa Matching MC data setb

Scanner kVp Head Body Head Body

G1 120 0.91 0.71 20�13� 5 �5�G1 140 0.93 – 20�13� –�20�G2 120 0.89 0.72 20�13� 5�20�G2 140 0.91 0.80 20�13� 20�18�G3 120 0.80 0.57 8�10� 7�10�G4 120 0.78 0.66 11 �9� 11�11�G5 120 0.84 0.64 9 �8� 8 �8�G5 140 0.89 0.73 20 �9� 5 �5�S1 120 1.07 1.17 12�12� 16�16�S1 140 1.09 1.27 12�12� 1 �1�S2 110 0.95 0.79 17�17� 20�20�S2 130 0.98 0.90 22�22� 18�18�P1 130 0.71 0.82 10 �6� 20 �5�P1 140 0.71 0.88 10�10� 22�20�H1 120 0.96 0.97 17– 19–

aImPACT factors �ImF� are derived from Eq. �8� and used to match scannersin Monte Carlo data sets.bNumbers are the matched MC data sets �MCSET� in this study and theImPACT database �in parentheses� for each scanner with available tubevoltages.

cans
ng; �b� ImPACT factors of body scanning.


set 7 and 10 for scanner G3, is about 28% for routine chestexaminations with the same setting of technical factors.However, the variations of effective doses for other scannerswith different MC data sets are all within 10% for routinehead or chest examinations. Organ doses were obtained froman ImPACT calculator and NRPB-SR25011,12 by giving ei-ther measured free-in-air axial doses for scanners or Im-PACT database. Organ doses were then used to calculateeffective dose by Eq. �7�. Table VII provides the mean valuesand standard deviations of organ doses and effective dosesfor several CT examinations assessed in this study based onthe surveyed data. Since the red bone marrow �RBM� and thegonads are the organs or tissues of special interest in radia-tion protection, their doses are given. The doses to the ova-ries and the testes during the pelvis examination deservemore attention. Table VII also shows the effective doses forfemale, male, and the average of both. The effective dosesare equal for females and males during the head and chestexaminations, but higher for females during scanning of theabdominal and pelvic regions.

Figure 5 plots the distribution of effective doses calcu-lated using the surveyed data as a function of DLP for theroutine head examinations. An approximate straight line�solid line� seems to fit the data quite well. The deviation of

TABLE VII. Organ doses and effective doses for CT exvalues ± standard deviations.

Examination

Organ doses �mSv

RBMa Ovaries

Routine Head 2.5 ± 1.5 –Routine Chest 6.4 ± 5.2 –Routine Abdomen 5.2 ± 4.3 4.5 ± 6.7Routine Pelvis 6.0 ± 4.7 23 ± 17LSPb,c 8.1 ± 6.7 2.4 ± 2.7Kidneysc 6.5 ± 6.2 3.2 ± 6.6

aRBM: red bone marrow.bLSP: liver-spleen-pancreas.cData in LSP and kidneys examinations are the resultsequence according to the protocol in each hospital.

FIG. 5. A plot of the distribution of effective dose as a function of DLP forthe head examination. The resulting correlation coefficient turned out to behigh �R2=0.97� and the region-specific normalized effective dose �EDPL� is
0.024.

data from the fitting line is small at low DLP but increaseswith DLP. The significant correlation for the examinations ofthe head �R2=0.97�, chest �R2=0.96�, abdomen �R2=0.99�,pelvis �R2=0.98�, liver-spleen-pancreas �R2=0.98�, and kid-neys �R2=0.99� indicates that the DLP is a good descriptor toestimate the effective dose by Eq. �9�. Table VIII shows acomparison of the effective dose per DLP, i.e., EDLP, ob-tained in this study, Italy, and the EC. It reveals that the EDLP

is much smaller for the head examination than other exami-nations. The highest percentage difference between calcula-tions and fittings is 9.4% for the abdomen examinations inthis study, but −7.9% for routine pelvic studies in Italy. Com-parison of the EDLP in this study, Italy and the EC in TableVIII shows good agreement within 10%.

Table IX gives the mean values and the third quartilevalues of CTDIw, DLP and effective doses determined in thisstudy and published elsewhere. The present values of CTDIw

are close to other published data, but the mean value for thehead examination is higher than Wales and Tanzania, butlower than Italy and Greece. Although the mean values andthird quartile values of DLP are all lower than or close to thedata of Italy, Greece, and Wales, the DLP values for the chestexamination are higher than the Greek data because the meanscan range in Taiwan �22.1 cm� is longer than in Greece�20.7 cm�. Similar results are shown for effective doses. All

ations calculated in the present work. Data are mean

Effective dose �mSv�

estes Female Male Average

– – – 1.6 ± 0.9– – – 8.4 ± 6.5± 0.1 7.9 ± 6.2 7.0 ± 5.3 7.4 ± 5.8± 3.2 9.6 ± 7.0 5.8 ±4.3 7.7 ± 5.7± 0.8 13 ± 11 13 ± 11 13 ± 11± 0.3 11 ± 10 11 ± 9.4 11 ± 9.6

complete examinations with one or more than one

TABLE VIII. Comparison of the EDLP in this study, Italy, and the EC.

Examination

This study Italy �Ref. 23� EC �Ref. 7�

EDLP R2 EDLPa R2 EDLP

Routine head 0.0024 0.97 0.0024 0.97 0.0023Routine chest 0.0176 0.96 0.0163 0.99 0.0170Routine abdomen 0.0164 0.99 0.0149 0.99 0.0150Routine pelvis 0.0183 0.98 0.0175 1.00 0.0190LSPb 0.0172 0.98 – –Kidneys 0.0171 0.99 – –

aEDLP of Italy in Ref. 23 terms as the angular coefficient.b

amin

�

T

0.13.60.10.1

s from

LSP: liver, spleen, and pancreas.


the values in the present study are lower than the EC, exceptthe third quartile value of CTDIw for routine head examina-tions, which is slightly higher.

IV. CONCLUSIONS

A national survey including on-site measurements andquestionnaires was conducted to establish the dose guidancelevels of CT in Taiwan. The x ray beam quality and outputand phantom doses were measured for nine representativeCT scanners. The collimation factor and the ImPACT factorof each representative scanner were determined from mea-sured data. Questionnaires were collected from CT facilitiesof 146 scanners utilizing 20 models from six manufacturers.Information on patient, procedure, scanner, and technique forthe head and body examinations and on CT quality controlprogram was provided. The CTDIw, the DLP, organ doses,and effective dose were calculated by using measured data,questionnaire information, and MC simulation results.

The present results showed that CT dose assessmentswithout considering the collimation factor could underesti-mate more than two times for the narrowest beam. Similarassessments applying the ImPACT database could possiblygenerate 28% uncertainty, as compared to those applying themeasured ImPACT factor data. Therefore, it is suggested toinclude the collimation factor25 and to apply the measuredImPACT factor in the estimation of CT patient doses. Fur-ther, it revealed that values of CTDIw and DLP for a givenexamination varied significantly among hospitals. This indi-cated that CT dose guidance levels were required in order to

TABLE IX. Comparison of present results on CTDIw,routine CT examinations. Data are mean values. Datindicates the total CT scanners or facilities studied.

Taiwan Italya Greecn 146 48 14

Head 55 �62� 60 �69� 68 �7Chest 20 �21� 20 �25� 21 �2Abdomen 22 �23� 24 �26� 23 �2Pelvis 22 �23� 25 �29� 27 �3

Head 665 �763� 725 �915� 919 �1Chest 455 �535� 473 �627� 429 �5Abdomen 453 �500� 517 �602� 493 �4Pelvis 410 �459� 467�501� 538 �6

E

Head 1.6 �1.7� 1.7 �2.1� 2.1 �2Chest 8.4 �9.7� 8.0 �10.7� 7.3 �8Abdomen 7.4 �8.2� 7.8 �9.1� 7.4 �7Pelvis 7.6 �8.3� 8.9 �9.5� 10.3 �1

aReference 23.bReference 27.cReference 28.

reduce patient doses to as low as reasonably achievable


�ALARA�. Dose guidance levels were derived using a cost-effective analysis on CTDIw and DLP. Results were given tothe national authorities as references in developing the qual-ity control standards and protocols.

Although it has been questioned the utility of CTDI forpatient dose,26 the various forms of CTDI were widelyadopted as the dose descriptor in CT. Measurements of CTDIare not only adequate for comparisons of scanner output suchas CTDIair and CTDIw, but also useful to check whether se-lected technical parameters are proper. Also, patient effectivedose could rapidly be assessed from DLP, derived fromCTDIw, shown on the control panel for operators. Finally,this survey should be useful as a reference baseline for fur-ther assessing doses contributed from multislice CT.

ACKNOWLEDGMENTS

This research was supported by the National ScienceCouncil of the Republic of China. The authors would like toacknowledge all the hospitals which returned completedquestionnaires and some permitted measurements on theirscanners. The authors also thank Wei-Chun Leon Li andChen-Cheng Liu for the assistance in measurements. H.Y.T.is grateful to Dr. Szu-Li Chang for encouragement and help-ful discussions.

a�Author to whom correspondence should be addressed. Electronic mail:[email protected]

1United Nations Scientific Committee on the Effects of Atomic Radiation“Sources and effects of ionizing radiation,” UNSCEAR 2000 report to theGeneral Assembly, with scientific annexes, Vol. I: Sources, Annex D

and effective dose with previously published data forarentheses are third quartile values. The number �n�

Walesc Norwayd Tanzaniae ECf

18 49 8

DIw �mGy�

46 43 �54� 6017 17 �18� 3022 17 �18� 3523 19 �23� 35

P �mGy cm�

731 913 �908� 1050663 783 �860� 650745 982 �1083� 780646 908 �1127� 570

ive dose �mSv�

2.0 2.1 �2.1� 2.411.5 13 �15� 11.712.8 15 �16� 11.19.8 17 �22� 10.8

dReference 29.eReference 30.fReference 7.

DLPa in p

eb

CT

8�5�2�0�

DL

093�03�93�02�ffect

.5�

.5�

.4�1.8�

�United Nations, New York, 2000�.


2National Radiological Protection Board, “Guidelines on patient dose topromote the optimization of protection for diagnostic medical exposures,”Vol. 10, Report No. 1 �Chilton, 1999�.

3International Commission on Radiological Protection, “Radiological pro-tection and safety in medicine,” Publication 73 �Pergamon Press, Oxford,1996�.

4International Commission on Radiological Protection, “Radiation andyour patient: A guide for medical practitioners,” Publication 88 �Perga-mon Press, Oxford, 2001�.

5International Commission on Radiological Protection, “Managing patientdose in computed tomography,” Publication 87 �Pergamon Press, Oxford,2001�.

6International Atomic Energy Agency, “International basic safety stan-dards for protection against ionizing radiation and for the safety of radia-tion sources,” IAEA Safety Series No. 115 �Vienna, Austria, 1996�.

7European Communities, “European guidelines on quality criteria for com-puted tomography,” EUR 16262 �Luxembourg, 1999�.

8European Communities, “European guidelines of quality criteria for di-agnostic radiographic images,” EUR 16260 �Luxembourg, 1996�.

9European Communities, “Guidance on diagnostic reference levels formedical exposures,” Radiation Protection 109 �Luxembourg, 1999�.

10M. A. Lewis, S. Edyvean, S. A. Sassi, H. Kiremidjian, N. Keat, and A. J.Britten, “Estimating patient dose on current CT scanners: Results of theImPACT CT dose survey,” available at: http://www.impactscan.org/dosesurveysummary.htm, accessed 17 July 2006.

11D. G. Jones and P. C. Shrimpton, National Radiological Protection Board,“Survey of CT practice in the UK, Part 3: Normalized organ doses cal-culated using Monte Carlo techniques,” NRPB-R250 �Chilton, 1991�.

12D. G. Jones and P. C. Shrimpton, National Radiological Protection Board,“Normalized organ doses for x-ray computed tomography calculated us-ing Monte Carlo techniques,” NRPB-SR250 �Chilton, 1993�.

13ImPACT, “ImPACT CT patient dosimetry calculator,” version 0.99x �Im-aging Performance Assessment of CT Scanners, London, 2006�.

14C. J. Tung, H. Y. Tsai, S. H. Lo, C. N. Guan, and Y. B. Chen, “Determi-nation of guidance levels of dose for diagnostic radiography in Taiwan,”Med. Phys. 28, 850–857 �2001�.

15S. Edyvean, M. A. Lewis, and A. J. Britten, “Imaging performance as-sessment of CT scanners and St. George’s Healthcare, “CT Scanner DoseSurvey: Measurement Protocol” �London, 1997�.

16American Association of Physicists in Medicine, “Standardized methodsfor measuring diagnostic x-ray exposures,” Report No. 31 �New York,NY, 1990�.

17P. C. Shrimpton, D. Hart, M. C. Hillier, B. F. Wall, and K. Faulkner,National Radiological Protection Board, “Survey of CT practice in theUK, Part 1: Aspects of examination frequency and quality assurance,”
NRPB-R248 �Chilton, 1991�.

18P. C. Shrimpton, D. G. Jones, M. C. Hillier, B. F. Wall, J. C. Le Heron,and K. Faulkner, National Radiological Protection Board, “Survey of CTpractice in the UK, Part 2: Dosimetric aspects,” NRPB-R249 �Chilton,1991�.

19International Electrotechnical Commission, “Evaluation and routine test-ing in medical imaging departments, Part 2-6: Constancy tests-x-rayequipment for computed tomography,” 61223-2-6 �Geneva, Switzerland,1994�.

20IPSM, Institute of Physical Sciences in Medicine, “National protocol forpatient dose measurements in diagnostic radiology,” �York, UK, 1992�.

21International Commission on Radiological Protection, “1990 recommen-dations of the International Commission on Radiological Protection,”Publication 60 �Pergamon Press, Oxford, 1991�.

22S. Edyvean, “CT dosimetry and a data base for CTDI values,” availableat: http://www.impactscan.org/slides/ecr2002/ctdosimetrydatabase.pdf,accessed Jan. 11, 2007.

23D. Origgi, S. Vigorito, G. Villa, M. Bellomi, and G. Tosi, “Survey ofcomputed tomography techniques and absorbed dose in Italian hospitals:A comparison between two methods to estimate dose and to verify fulfil-ment of the diagnostic reference levels,” Eur. Radiol. 16, 227–237�2006�.

24H. Y. Tsai, Radiation Dosimetry of Computed Tomography �Ph.D. disser-tation, Department of Nuclear Science, National Tsing Hua University,Hsinchu, Taiwan, 2003�.

25N. Theocharopoulos, J. Damilakis, K. Perisinakis, and A. Tzedakis, “Es-timation of effective doses to adult and pediatric patients from multislicecomputed tomography: A method based on energy imparted,” Med. Phys.33, 3846–3856 �2006�.

26D. J. Brenner, C. H. McCollough, and C. G. Orton, “It is time to retire thecomputed tomography dose index �CTDI� for CT quality assurance anddose optimization,” Med. Phys. 33, 1189–1191 �2006�.

27D. Papadimitriou, A. Perris, A. Manetou, M. Molfetas, N. Panagiotakis,M. Lyra-Georgosopoulou, K. Hourdakis, S. Kottou, G. Tosi, D. Origgi,and S. Vigorito, “A survey of 14 computed tomography scanners inGreece and 32 scanners in Italy. Examination frequencies, dose referencevalues, effective doses and doses to organs,” Radiat. Prot. Dosim. 104,47–53 �2003�.

28P. A. Hiles, S. E. Brennen, S. A. Scott, and J. H. Davies, “A survey ofpatient dose and image quality for computed tomography scanners inWales,” J. Radiol. Prot. 21, 345–354 �2001�.

29H. M. Olerod, “Analysis of factors influencing patient doses from CT inNorway,” Radiat. Prot. Dosim. 71, 123–133 �1997�.

30J. E. Ngaile, P. Msaki, and R. Kazema, “Towards establishment of thenational reference dose levels from computed tomography examinations
in Tanzania,” J. Radiol. Prot. 26, 213–225 �2006�.
http://dx.doi.org/10.1118/1.1368126

http://dx.doi.org/10.1118/1.2349694

http://dx.doi.org/10.1118/1.2173933

http://dx.doi.org/10.1088/0952-4746/21/4/302

http://dx.doi.org/10.1088/0952-4746/26/2/006

survey of computed tomography scanners in taiwan: dose descriptors, dose guidance levels, and...

Documents