Technical assessment of a cone-beam CT scanner for otolaryngologyimaging: Image quality, dose, and technique protocols

J. XuDepartment of Biomedical Engineering, Johns Hopkins University, Baltimore, Maryland 21205

D. D. Reh and J. P. CareyDepartment of Otolaryngology – Head and Neck Surgery, Johns Hopkins University,Baltimore, Maryland 21205

M. MaheshRussell H. Morgan Department of Radiology, Johns Hopkins University, Baltimore, Maryland 21287

J. H. Siewerdsena)

Department of Biomedical Engineering, Johns Hopkins University, Baltimore, Maryland 21205 and Russell H.Morgan Department of Radiology, Johns Hopkins University, Baltimore, Maryland 21287

(Received 15 March 2012; revised 8 June 2012; accepted for publication 13 June 2012; published 25July 2012)

Purpose: As cone-beam CT (CBCT) systems dedicated to various imaging specialties proliferate,technical assessment grounded in imaging physics is important to ensuring that image quality andradiation dose are quantified, understood, and justified. This paper involves technical assessment of anew CBCT scanner (CS 9300, Carestream Health, Rochester, NY) dedicated to imaging of the ear andsinuses for applications in otolaryngology-head and neck surgery (OHNS). The results guided eval-uation of technique protocols to minimize radiation dose in a manner sufficient for OHNS imagingtasks.Methods: The technical assessment focused on the imaging performance and radiation dose for eachof seven technique protocols recommended by the manufacturer: three sinus protocols and four ear(temporal bone) protocols. Absolute dose was measured using techniques adapted from AAPM TaskGroup Report No. 111, involving three stacked 16 cm diameter acrylic cylinders (CTDI phantoms)and a 0.6 cm3 Farmer ionization chamber to measure central and peripheral dose. The central dose(Do) was also measured as a function of longitudinal position (z) within and beyond the primaryradiation field to assess, for example, out-of-field dose to the neck. Signal-difference-to-noiseratio (SDNR) and Hounsfield unit (HU) accuracy were assessed in a commercially availablequality assurance phantom (CATPHAN module CTP404, The Phantom Laboratory, Greenwich,NY) and a custom phantom with soft-tissue-simulating plastic inserts (Gammex RMI, Madison,WI). Spatial resolution was assessed both qualitatively (a line-pair pattern, CATPHAN moduleCTP528) and quantitatively (modulation transfer function, MTF, measured with a wire phantom).Imaging performance pertinent to various OHNS imaging tasks was qualitatively assessed using ananthropomorphic phantom as evaluated by two experienced OHNS specialists.Results: The technical assessment motivated a variety of modifications to the manufacturer-specifiedprotocols to provide reduced radiation dose without compromising pertinent task-based imaging per-formance. The revised protocols yielded Do ranging 2.9–5.7 mGy, representing a ∼30% reduction indose from the original technique chart. Out-of-field dose was ∼10% of Do at a distance of ∼8 cmfrom the field edge. Soft-tissue contrast resolution was fairly limited (water-brain SDNR ∼0.4–0.7)while high-contrast performance was reasonably good (SDNR ∼2–4 for a polystyrene insert in theCATPHAN). The scanner does not demonstrate (or claim to provide) accurate HU and exhibits asystematic error in CT number that could potentially be addressed by further calibration. The spa-tial resolution is ∼10–16 lp/cm as assessed in a line-pair phantom, with MTF exceeding 10% out to∼20 lp/cm. Qualitative assessment by expert readers suggested limited soft-tissue visibility but excel-lent high-contrast (bone) visualization with isotropic spatial resolution suitable to a broad spectrumof pertinent sinus and temporal bone imaging tasks.Conclusions: The CBCT scanner provided spatial and contrast resolution suitable to visualizationof high-contrast morphology in sinus, maxillofacial, and otologic imaging applications. Rigoroustechnical assessment guided revision of technique protocols to reduce radiation dose while main-taining image quality sufficient for pertinent imaging tasks. The scanner appears well suited to

4932 Med. Phys. 39 (8), August 2012 © 2012 Am. Assoc. Phys. Med. 49320094-2405/2012/39(8)/4932/11/$30.00

4933 Xu et al.: Technical assessment of a CBCT scanner for otolaryngology imaging 4933

high-contrast sinus and temporal bone imaging at doses comparable to or less than that reported forconventional diagnostic CT of the head. © 2012 American Association of Physicists in Medicine.[http://dx.doi.org/10.1118/1.4736805]

Key words: cone-beam CT, image quality, radiation dose, technical assessment, otolaryngology, ENTimaging, maxillofacial imaging, temporal bone imaging

I. INTRODUCTION

X-ray computed tomography (CT) has proliferated overthe last several decades as an important medical imagingmodality with widespread application in diagnosis, surgicalguidance, and monitoring. Chief among the considerationsin applying this prevalent modality are radiation dose andimaging performance. As multidetector CT (MDCT) contin-ues to grow and benefit from appropriate utilization criteria,low-dose scan protocols, iterative reconstruction techniques,multienergy capabilities, and new applications, a variety ofapplication-specific embodiments of cone-beam CT (CBCT)have emerged over the last decade. Most current embodimentsof CBCT employ variations of the Feldkamp-Davis-Kress(FDK) algorithm1 for 3D filtered backprojection, althoughCBCT will similarly benefit from advances in iterative re-construction techniques and low-dose protocols.2 The scopeof application-specific CBCT embodiments include den-tal/maxillofacial imaging,3–13 temporal bone imaging,14–16

breast imaging,17 musculoskeletal imaging,18 image-guidedradiotherapy,19, 20 and image-guided surgery.21, 22 Each offersthe potential for nearly isotropic, submillimeter spatialresolution combined with soft-tissue contrast resolution ap-proaching that of MDCT (but typically limited by factors suchas x-ray scatter, limited field of view (FOV), and increasednoise).

The proliferation of these new applications of CBCT—each involving new system geometries, scan orbits, radiationdose profiles, and image quality characteristics—heightensthe need for quantitative technical assessment grounded inscientific methodology and imaging physics to rigorouslyquantify the performance of such systems, ensure that theyare appropriately deployed, understand their performance ca-pabilities with respect to specific imaging tasks, and guideknowledgeable selection of technique protocols. This pa-per concerns the technical assessment of a new CBCT sys-tem (CS 9300, Carestream Health, Rochester, NY) devel-oped specifically for otolaryngology-head and neck surgery(OHNS) and maxillofacial imaging [alternatively – ear, nose,and throat (ENT) imaging]. Comparable systems now com-mercially available in a range of dental/ENT applications in-clude the MiniCAT (Xoran, Ann Arbor, MI), CB Mercuray(Hitachi, Twinsburg, OH), NewTom (QR, Verona, Italy), i-CAT (Imaging Sciences, Hatfield, PA), Accuitomo 170 (J.Morita USA, Irvine, CA), and others. The CS 9300 includesmodifications of various characteristics in comparison to pre-vious platforms (9000 series) from the same manufacturer,including: options for expanded field of view (FOV); a va-riety of full scan (360◦) and short scan (at least 180◦ + fanangle) protocols with various FOV and scan speed; pulsed or

continuous x-ray source; and enhanced acquisition and recon-struction software.

The technical assessment reported below addresses thedosimetric properties and image quality associated withmanufacturer-specified technique protocols of the CS 9300for OHNS imaging. Results are compared to reports in thescientific literature, but a head-to-head comparison of per-formance versus other CBCT (or MDCT) systems15, 23, 24

is beyond the objectives of the current work. Rather, thework reported below focuses on a system-specific techni-cal assessment that was performed, first, to guide knowl-edgeable selection of minimum-dose protocols sufficientfor relevant imaging tasks prior to deployment in clinicalstudies at our institution, and second, as a basis of per-formance comparison with other systems reported in theliterature.

II. METHODS AND MATERIALS

II.A. The CS 9300 and default protocols

According to the manufacturer, the intended use of the CS9300 scanner is “. . . to produce 3D digital x-ray images ofthe dento-maxillo-facial and ENT regions as diagnostic sup-port for pediatric and adult patients.” The scanner capabili-ties and specifications are summarized in Table I. The defaultimaging protocols deployed on the system are summarized inTable II, including three sinus protocols (denoted S) and fourear (temporal bone) protocols (denoted E). All measurementsinvolving the unilateral temporal bone protocols (E2, E3, andE4) were performed with the right ear protocol. The left earprotocols were spot-checked to be symmetric with respect tothe contralateral side. The angular extent, FOV, and number ofprojections are all nonmodifiable parameters for each imag-ing protocol, but the kVp and mAs may be freely adjusted atthe discretion of the technologist. The “short-scan” orbits arecomparable to half-scan orbits (180◦ + fan), with orbital ex-tent particular to each protocol detailed separately below andin Table II.

An initial technical assessment was performed (data notshown) that motivated modifications to the scan orbits (startand stop angles of the x-ray source and detector) and the tech-nique chart (reduction in kVp and mAs). Results presentedbelow pertain to protocols and measurements after such mod-ifications unless specifically noted. The resulting protocols(detailed below) reduced dose by up to 30% and better sit-uated the short-scan orbits to impart dose preferentially to theposterior of the head (and reduce anterior dose – e.g., to theeyes).

Medical Physics, Vol. 39, No. 8, August 2012

4934 Xu et al.: Technical assessment of a CBCT scanner for otolaryngology imaging 4934

TABLE I. Summary of system parameters and specifications. The x-ray tubeis the CEI OPX 110 (Trophy, Verona, Italy), and the x-ray detector is Model2520 (Varian Imaging Products, Palo Alto, CA).

Parameter Value

X-ray tube CEI OPX 110Power (max) 1.5 kWAnode Fixed target (W)Focal spot size 0.7 mmTube voltage 60–90 kVp (1 kVp steps)Tube current 2–15 mA (1 mA steps)X-ray pulse rate 33 p/s (1 pulse per 30 ms)

Inherent filtration 2.5 mm Al equiv. (70 kVp)Added filtration 0.1 mm Cu (70 kVp)Bowtie filter Custom (Cu)HVL (70 kVp) 4.6 mm AlHVL (75 kVp) 5.0 mm AlHVL (80 kVp) 5.4 mm AlHVL (85 kVp) 5.7 mm AlHVL (90 kVp) 6.0 mm Al

Detector type Varian 2520Detector readout mode Dynamic gainPixel size (intrinsic) 0.127 mmPixel binning

(S1, S3, S3, E1, E4) 2 × 2 (0.254 mm)(E2, E3) 1 × 1 (0.127 mm)

X-ray converter CsI:TlAntiscatter grid None

Reconstruction filter Not specifiedVoxel Size 90–500 μm

II.B. Dose measurement: Experimental setup

Dose measurements were performed with methodologyadapted from those outlined in AAPM Task Group ReportNo. 111.25 As shown in Fig. 1, three acrylic cylindrical phan-toms of 16 cm diameter (CTDI phantoms) were stacked alongthe central longitudinal axis to simulate the “head,” and a0.6 cm3 Farmer ionization chamber was used in conjunctionwith a Radcal electrometer (AccuDose, Radcal Corp., Mon-rovia, CA) to measure the central and peripheral doses im-parted for all protocols listed in Table III. Dose measure-

Xray Tube

Detector

CTDIPhantoms

Temple Wands

ControlConsole

FIG. 1. Experimental setup for dose measurements. The photograph showsthe scanner with the chin rest removed and a stack of 16 cm diameter plas-tic cylinder phantoms. Variations of the phantom setup included: three 16cm acrylic CTDI phantoms; a CATPHAN in place of the central cylinder(for SDNR measurements); a custom SolidWaterTM cylinder with tissue-simulating plastic inserts (Gammex RMI, Madison, WI); a wire phantom (forMTF measurements); and an anthropomorphic head phantom (natural skele-ton in RandoTM plastic).

ments used up-to-date manufacturer calibration of the elec-trometer and accounted for temperature-pressure correctionsat the time of measurement. Measurements were nominallyperformed at the level of the central axial slice of the im-age volume. A further measurement of dose as a function ofkVp and mAs was performed for the S1 protocol and the E1

TABLE II. Technique chart for various protocols deployed on the scanner. Three sinus protocols include: S1 (large FOV); S2 (fast scan, small FOV); and S3(small FOV, high quality). Four ear protocols include: E1 (bilateral FOV); E2 (unilateral (R or L), high-resolution); E3 (unilateral, fast); and E4 (unilateral,larger FOV). Each protocol entails different FOV, scan time, number of projections per scan, radiation dose, scan angle, and image quality. Voxel size (mm) isisotropic in x, y, and z directions. Finer voxel size (0.09 mm) for the E2-E4 protocols have since become available on the system but were not investigated in thecurrent study due to increased image noise.

Protocol kVp mA mAs FOV (cm3) Scan time (s) Orbital extent Voxel size (mm) Volume size (voxels)

Sinus 1 (S1) 85 5 56 17 × 13.5 × 17 11.3 Full (360o) 0.3 567 × 450 × 567Sinus 2 (S2) 85 5 32 17 × 11 × 17 6.4 Short (∼204◦) 0.5 339 × 220 × 339Sinus 3 (S3) 85 5 51 17 × 11 × 17 10.2 Short (∼204◦) 0.3 567 × 367 × 567Ear 1 (E1) 90 5 43 17 × 6 × 17 8.5 Short (∼204◦) 0.2 850 × 300 × 850Ear 2 (E2) 90 6.3 126 5 × 5 × 5 20 Short (∼188◦) 0.2 250 × 250 × 250Ear 3 (E3) 90 6.3 76 5 × 5 × 5 12 Short (∼188◦) 0.2 250 × 250 × 250Ear 4 (E4) 90 6.3 76 8 × 8 × 8 12 Short (∼192◦) 0.3 267 × 267 × 267

Medical Physics, Vol. 39, No. 8, August 2012

4935 Xu et al.: Technical assessment of a CBCT scanner for otolaryngology imaging 4935

TABLE III. Summary of dose measurements for each protocol deployed on the scanner. Central dose (Do), peripheral dose (P1–P4), and various aggregatecalculated dose values are shown. Labels for peripheral positions are provided for P1-P4, where A = anterior, P = posterior, R = Right, and L = left. Theprotocols are in Table II. The measurement locations are shown in Fig. 2.

Do Do/mAs P1 (A) P2 (R) P3 (P) P4 (L) D̄periph Dw Dw /mAs DLP DAP DE

# (mGy) (mGy/mAs) (mGy) (mGy) (mGy) (mGy) (mGy) (mGy) (mGy/mAs) (mGy cm) (mGy cm2) (mSv)

S1 5.7 0.10 6.1 6.1 6.1 6.1 6.10 5.97 0.11 80.6 1370 0.19S2 3.3 0.10 1.4 4.0 4.5 4.0 3.46 3.41 0.11 37.5 637 0.09S3 5.3 0.10 2.2 6.3 7.2 6.5 5.55 5.47 0.11 60.2 1023 0.14E1 4.5 0.11 1.6 5.5 8.3 5.7 5.29 5.02 0.12 30.1 512 0.07E2 4.8 0.04 6.4 2.5 12.1 8.2 7.30 6.46 0.05 32.3 162 0.07E3 2.9 0.04 3.8 1.5 7.2 4.0 4.12 3.70 0.05 18.5 92 0.04E4 3.2 0.04 5.5 2.1 6.2 5.7 4.88 4.31 0.05 34.5 276 0.08

protocol. The central dose (Do) was defined as the absolutedose (mGy) at the center of the CTDI phantom for each scanfor each protocol.

Peripheral dose was measured at four cardinal locations atthe periphery of the CTDI phantom (at the same level as thecentral dose), with all other experimental factors held con-stant. Since several of the protocols involved short-scan orbitsof the source and detector about the head, the peripheral dosevaried at each of the measurement points (e.g., highest atthe posterior point for short-scan orbits in which the sourcetraverses the posterior of the head). In addition to the centralabsolute dose (Do) the four peripheral dose measurements(Dperiph) were averaged to yield a “weighted” dose valueanalogous to CTDIw, specifically: (DW = 1

3Do + 23 D̄periph).

The dose-area product was given by DAP = DW · L · W ,where L and W refer to the length and width of the imageFOV. Similarly, the dose-length product was DLP = DW · L.To the limited extent that is meaningful to convert theabsolute dose from such orbits to the “effective dose” (DE,mSv) and to permit comparison to other systems for whichresults have been reported only in terms of effective dose(mSv), we used the tissue conversion factor for the head(kHead = 0.0023 mSv/mGy/cm) given by ICRP Publication10326 and computed DE = kHead · DLP. The limitation andapproximation of this approach is recognized—namely, thateffective dose conversion for short-scan orbits is not strictlydefined. Specifically, the required tissue conversion factorswere developed in the context of conversion from CTDIw.The short-scan measurements of absolute dose (mGy) arevalid, but the effective dose values (mSv) should be recog-nized as approximate and are only included for comparisonwith other systems that only report mSv.

Dose distribution “maps” were generated using asmoothed interpolation of the five measurement points (thecentral dose and four peripheral doses) for each protocol. Thedose maps provide visualization of heterogeneous dose distri-bution about the lateral, posterior, and anterior aspects of thehead, particularly for the various short-scan protocols. Theydo not pretend to account for tissue heterogeneities, thoughthey are a valuable means of conveying dose distributions tothe clinicians and manufacturer with respect to the variousscan orbit pathways.

A further study was conducted with S1 and E1 protocols tocharacterize the out-of-field dose [Do(z)] along the longitudi-nal axis. The absolute dose to the center of the CTDI phantomwas measured as a function of z (longitudinal position) begin-ning at the central plane, covering the extent of the primarycollimated x-ray field, and extending inferiorly beyond thefield toward the “neck.” The same experimental setup of threestacked CTDI phantoms was used for this assessment, withthe ionization chamber position manually translated along thez axis in ∼2 cm increments.

II.C. Imaging performance

Performance measurements used two CatPhan modules(CTP404 and CTP528) and a custom SolidWaterTM cylinderwith tissue-simulating plastic inserts (Gammex RMI, Madi-son, WI). The signal difference to noise ratio (SDNR) wascalculated as follows:

SDNR = 2|μ̄insert − μ̄background|σinsert + σbackground

where μ̄insert is the average voxel intensity of a specified in-sert, μ̄background is the average voxel intensity of the back-ground material adjacent to (and at the same radius as) theinsert, and σ insert and σ background are the standard deviations inthe respective regions.

II.C.1. High-contrast SDNR

The CTP404 insert containing various plastic cylindricalinserts was used to assess high-contrast SDNR. For all pro-tocols, the SDNR was calculated between polystyrene (mea-sured 91 HU) and background (measured −26 HU). Thedose-normalized SDNR was computed by dividing by thesquare root of the measured absolute dose (Do) for each pro-tocol. Due to the presence of a significant blush and ring ar-tifact and lateral truncation artifacts (depending on FOV), theregion of interest (ROI) location for calculation of SDNR wasselected to avoid such artifacts while maintaining equal radiusfrom the center of the image for all inserts and backgroundROIs.

Medical Physics, Vol. 39, No. 8, August 2012

4936 Xu et al.: Technical assessment of a CBCT scanner for otolaryngology imaging 4936

II.C.2. Low-contrast (Soft-tissue) SDNR

Further study of the low-contrast resolution capabilitiesof the scanner was performed using a SolidWater phan-tom with inserts that simulated soft-tissue densities. Tissue-equivalent inserts (Gammex RMI, Madison, WI) includedadipose (−112 HU), solid water (0 HU), brain (6 HU), andliver (87 HU). Soft-tissue SDNR was calculated in the samemanner as described above.

II.C.3. HU accuracy

The same phantoms were scanned with a MDCT scanner(Somatom Definition Flash, Siemens Healthcare, Forcheim,Germany) using standard clinical “head” protocols (120 kVp,125 mAs, T80f kernel, 0.4 × 0.4 × 0.4 mm3 voxel size), andthe HU values reported by the MDCT scanner and CS 9300scanner were compared.

II.C.4. Spatial resolution

Spatial resolution was assessed qualitatively using a line-pair phantom (CTP528 module of the CatPhan) for all proto-cols. Quantitative assessment of spatial resolution for the S1and E1 protocols was performed by measurement of the mod-ulation transfer function (MTF) from a wire phantom. TheMTF was calculated as the Fourier transform of an oversam-pled line-spread function (LSF) derived from Radon trans-form of axial images of the wire within a cylindrical phan-tom under tension, slightly angled to the longitudinal imageaxis. Radon transform, oversampling, LSF normalization, andMTF estimation followed similar procedures as previouslypublished works.27, 28

II.C.5. Image quality in an anthropomorphichead phantom

An overall qualitative assessment of image quality wasperformed using scans of an anthropomorphic RANDO headphantom (natural human skeleton in tissue-equivalent plas-tic; The Phantom Laboratory, Greenwich, NY). Images werequalitatively assessed by a rhinologist and an otologist withrespect to the visibility of pertinent anatomical structures andoverall diagnostic quality. The potential for more quantitativeobserver performance assessment is recognized (e.g., ROCtests), but is beyond the scope of the technical assessment re-ported here. The qualitative interpretation by expert clinicianswas valuable, complementary, and confirmatory of measure-ments of SDNR and MTF.

III. RESULTS

The results detailed below correspond to a secondtechnical assessment of the CS 9300 after modificationswere made based upon recommendations arising froman initial technical assessment performed using the samemethods and experimental setup. A summary of resultsfrom the initial assessment for purposes of comparison isas follows. The initial sinus protocols (S1, S2, and S3)employed a 90 kVp beam (5 kVp greater than those listed in

Table II) and a tube current of 6.3 mA (1.3 mA higher thanthose listed in Table II). The ear protocols (E1, E2, E3, andE4) did not change in beam energy, but the mA was reducedfrom 6.3 mA to 5 mA for E1 and from 8 mA to 6.3 mAfor E2, E3, and E4. Several of the source-detector orbits inthe initial protocols were also modified to those illustratedin Fig. 2: The S2 protocol, for example, initially involved alonger arc beginning at the right ear, traversing the posteriorof the head, and ending anterior to the left ear; similarly,the E1 protocol involved an arc beginning posterior to theright ear, traversing the posterior of the head and ending atthe left anterior of the head; other scan trajectories were asshown in Fig. 2. The adjustment of the S2 and E1 protocolsto those shown in Fig. 2 was motivated primarily to reducethe total arc length and deposit dose posteriorly [rather thanto the anterior head (viz., eye lens)]. The initial assessmentyielded dose values of: Dw = 9.2 mGy (S1), 5.3 mGy (S2),8.5 mGy (S3), 6.5 mGy (E1), 7.9 mGy (E2), 5.0 mGy(E3), and 8.1 mGy (E4). Similarly, the SDNR/

√mGy from

the original assessment for the same CatPhan module was0.97 (S1), 1.2 (S2), 1.0 (S3), 1.2 (E1), 1.5 (E2), 1.7 (E3),and 0.75 (E4) [all units /sqrt(mGy)]. Spatial resolutionassessed subjectively from the CatPhan line-pair patternranged between 12 lp/mm (S2) and greater than 15 lp/mm(E2–E4). The reduction in kVp and mAs and modification ofthe source-detector orbits were qualitatively assessed in an-thropomorphic phantoms, with further possible modificationssuggested below.

III.A. Dose

Dose measurements demonstrated that all protocolsdeployed on the CS 9300 scanner are similar to (or somewhatlower than) those reported for comparable CBCT scannersas well as those reported for MDCT head protocols. Table IIIsummarizes the dose measurements, where central dose isseen to be in the range 2.9–5.7 mGy, depending on the specificprotocol. For example, the lowest and highest dose protocols(E3 and S1, respectively) gave Do = 2.9 and 5.7 mGy,Dw = 3.7 and 6.0 mGy, DLP = 18 and 81 mGy cm, DAP= 92 and 1370 mGy cm2, and DE = 0.04 and 0.19 mSv,respectively. By comparison, Ludlow et al.23 reported dosesfor comparable scanners to be between 0.05 mSv (NewTom)and 1 mSv (Mercury). Because other reports in the literatureutilized thermoluminescent dosemeters (TLDs) and wereaimed at patient dose characterization instead of absolutedose (“output”) of the scanner itself, a comparison in termsof absolute dose (mGy) is not available. In comparison toMDCT of the head, the median value for CTDIW reported byPantos et al.29 was approximately 52 mGy, with a range of17–181 mGy measured over 17 studies spanning approxi-mately two decades. The measured value of DW (the approxi-mate analog of CTDIW for short-scan orbits) for the CS 9300is considerably lower than the lowest CTDIW reported forMDCT, although ongoing advances in dose reduction andimproved reconstruction algorithms will likely drive MDCTto still lower values. Some of those same advances will likelyapply to further dose reduction in CBCT as well.

Medical Physics, Vol. 39, No. 8, August 2012

4937 Xu et al.: Technical assessment of a CBCT scanner for otolaryngology imaging 4937

2

3

4

5

6

7

8

3456789101112

2

3

4

5

6

7

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

θ

R L

A

P

TubeStart

TubeStop

Center ofObject

Center ofRotation

Key S1 S2 S3

E1 E2 E3 E4

5.0

5.2

5.4

5.6

5.8

6.0

6.2

6.4

1.5

2.0

2.5

3.0

3.5

4.0

4.5

2

3

4

5

6

7

FIG. 2. Dose distributions (“maps”) in the central axial plane for various scanner orbits. The colorbars show the dose (mGy). Each protocol is labeled as inTable I. The top left image “Key” shows the legend for: tube start angle, tube stop angle, center of rotation, and center of the object. The small FOV for theunilateral scan protocols are shown as dotted circles in E2, E3, and E4.

As shown in Fig. 2, the various scan orbits impart verydifferent dose distributions: for a 360◦ orbit (S1), the dose de-position is the expected, radially symmetric dose distributionwith exponential attenuation toward the center of the phan-tom; for the short-scan and unilateral orbits, however, the doseis deposited predominantly at the posterior of the head (S2,S3, and E1) and/or unilaterally (E2, E3, and E4). These dif-ferences in scan orbit yield variation in the peripheral dosethat in turn affects the “weighted” and “effective” dose shownin Table III. The dose maps in Fig. 2 further demonstrate thatthe short-scan orbits achieve considerable dose sparing of theanterior region, including the eyes. Some implications withrespect to dosimetry standards and further improvements inanterior dose sparing are discussed below.

Figure 3 shows measurements of the longitudinal (z) dis-tribution of dose within and beyond the primary beam field of

0.01

0.1

1

10

-10 -5 0 5 10 15 20 25

Mea

sure

d D

ose

at A

xis

of R

otat

ion

(mG

y)

Δz = Position of Measurement Along Axis (cm)

Sinus 1

Ear 1

13.5 cm

6 cm

FIG. 3. Out-of-field (longitudinal) dose profiles for the S1 and E1 protocols.The longitudinal dose tails fall off exponentially with z outside of the primarybeam. The range denoted by dotted lines labeled 13.5 cm and 6 cm denotethe estimated beam heights for protocol S1 and E1, respectively. These areprovided as a locational reference for the out-of-field radiation.

view. The falloff was anticipated to be fairly gradual due toscatter in the broad volumetric beam. The longitudinal dosetails fall to ∼10% of the maximum central dose at ∼8 cm fromthe edge of the FOV and to ∼1% of the maximum central doseat ∼12 cm from the edge of the FOV. Assuming an approxi-mate thyroid position at ∼8 cm below the chin, the dose to thethryoid would be approximately 1.1 mGy and 0.8 mGy for theS1 and E1 protocols, respectively. Previous work30 shows thata majority of the out-of-field dose arises from internal scatterthrough the patient, and a thyroid shield would not be effec-tive in limiting dose to thyroid, since x-ray scatter originatesin the head and travels “down” the neck.

III.B. Signal difference to noise ratio andCT number accuracy

The SDNR was measured for all sinus and temporal boneprotocols as summarized in Fig. 4. Overall, the temporal boneprotocols provided slightly improved SDNR in comparison tothe sinus protocols, attributed primarily to the smaller FOVand beam width, resulting in reduced x-ray scatter. The im-ages also illustrate the variation in FOV size and placement(shifts of the gantry as noted in Fig. 2): S1 covers the en-tire phantom; S2 and S3 shift the FOV anteriorly (to coverthe sinuses); E1 is intended to cover the bilateral posterior as-pect of the head (temporal bones); and E2, E3, and E4 placea smaller FOV unilaterally (L or R temporal bone). For theresults in Fig. 4, the phantom was not moved in cases S1, S2,S3, and E1, but was rotated in cases E2, E3, and E4 such thatthe polystyrene insert (indicated by the arrow in Fig. 4(a)) re-mained in the FOV for purposes of comparison and SDNRanalysis.

The low-contrast imaging performance was investigatedfurther for the S1 and E1 protocols to assess the potential forsoft-tissue visualization (beyond the fairly high-contrast in-serts of the CatPhan modules). Images of the 16 cm SolidWa-ter phantom with various tissue-equivalent inserts are shownin Fig. 5. Soft-tissue inserts include (W) solid water, (L)

Medical Physics, Vol. 39, No. 8, August 2012

4938 Xu et al.: Technical assessment of a CBCT scanner for otolaryngology imaging 4938

400

600

800

1000

1200

1400

Key S1 S2 S3

E1 E2 E3 E4

Polystyrene

Acrylic

Air

Delrin

Teflon

Air

PMP

LDPE

AcrylicSpheres

(b)

(a)

(c)

0

1

2

3

4

S1 S2 S3 E1 E2 E3 E4

SD

NR

Protocol #

0

0.5

1

1.5

2

S1 S2 S3 E1 E2 E3 E4

SD

NR

/√m

Gy

Protocol #

FIG. 4. Analysis of SDNR. (a) The CatPhan Module CTP404 was scanned at all sinus and ear protocols. An equivalent window and level was set for all cases,with the intensity grayscale (native scanner voxel values) shown to the right. (b and c) SDNR calculated for the polystyrene insert (indicated by the arrow in (a)).

liver, (B) brain, and (A) adipose. Apparent differences be-tween the two brain and liver inserts are due to variationsfrom the manufacturer. Only the liver (+87 HU) and adipose(−112 HU) inserts demonstrated a high level of conspicu-ity (SDNR ∼0.70 and 2.1, respectively). Soft-tissue visibility

was qualitatively inferior to the same object imaged in MDCT.A dark circular blush artifact is also evident, as is a de-gree of spatial nonuniformity (shading near the center, likelydue to x-ray scatter) that somewhat diminish overall imagequality.

Sinus 1 Ear 1

B

B

L

L

A

W

B

B

L

L

A

W

-500

-400

-300

-200

-100

0

FIG. 5. Soft-tissue phantom imaged using S1 and E1 protocols. The grayscale at right shows native voxel values reported by the scanner (not HU). Tissue-equivalent inserts are as follows: B (brain: 6 HU, SDNRS1 = 0.35, SDNRE1 = 0.7), W (water, 0 HU, SDNRS1 = 0.27, SDNRE1 = 0.20), L (liver, 87 HU,SDNRS1 = 0.70, SDNRE1 = 1.2), A (adipose, −112 HU, SDNRS1 = 2.1, SDNRE1 = 2.5), and SolidWater background (0 HU).

Medical Physics, Vol. 39, No. 8, August 2012

4939 Xu et al.: Technical assessment of a CBCT scanner for otolaryngology imaging 4939

y = 0.69x - 163

y = 0.79x - 166

- 1200

-900

-600

-300

0

300

600

900

1200

-1200 -800 -400 0 400 800 1200

Vox

el V

alue

(C

S930

0)

MDCT HU values

Sinus 1Ear 1

FIG. 6. Voxel values reported by the CS 9300 in comparison to HU valuesmeasured in MDCT (standard head protocol). As is common with CBCTsystems, voxel values exhibit a systematic error primarily associated withhigh x-ray scatter (i.e., slope less than unity and nonzero intercept). The lineof identity (bold) is shown as a basis of comparison.

Figure 6 shows the voxel values reported by the CS 9300plotted versus the HU reported by the MDCT scanner (stan-dard head protocol at 120 kVp, Siemens Somatom DefinitionFlash). Note that the manufacturer does not claim accurateHU calibration on the CS 9300. A fairly linear relationship isobserved, related by slopes of 0.7 and 0.8 for the S1 and E1protocols, respectively. A slope less than 1.0 is presumablyassociated with increased x-ray scatter (larger cone angle) forthe CS 9300, which appears to dominate over HU discrepan-cies associated with the lower kVp. The lower slope for the S1protocol compared to the E1 protocol is similarly consistentwith increased x-ray scatter associated with the larger FOV.This level of HU inaccuracy is typical for CBCT systems andpresent an area for further improvement through careful cali-bration procedures.31, 32

III.C. Spatial resolution

As shown in Fig. 7, all protocols exhibited spatial reso-lution better than 10 lp/cm in a qualitative assessment of theline-pair phantom. The unilateral temporal bone protocols(E2, E3, and E4) demonstrated the highest spatial resolution,∼13 lp/cm. The differences observed in the limiting spatialresolution among various protocols are attributed to thetechnique parameters shown in Table II, most notably voxelsize. Specifically, S1 and S3 (each with voxel size 0.3 mm)have superior spatial resolution compared to S2 (voxel size0.5 mm). The difference in spatial resolution between S1and S3 is more subtle and can be attributed to superiorview sampling for the latter – the number of views areapproximately equal for both protocols, but they are spreadover a larger angle for S1 than for S3.

More quantitative assessment of spatial resolution isshown in the MTF measurements of Fig. 8, where the S1 andE1 protocols were found to give MTF exceeding 10% out to20 lp/cm or more. The MTF is slightly improved for the E1protocol, owing to the smaller FOV (reduced x-ray scatter andfiner voxel sampling). The system interface in its current im-plementation does not allow adjustment of the reconstructionfilter (“kernel”), and the filters associated with each protocolare not reported. It is not known if the filter varies betweenprotocols.

III.D. Image quality assessed in an anthropomorphicRando phantom

Images of the anthropomorphic head phantom in Figs. 9and 10 provided qualitative assessment of the various scanprotocols with respect to pertinent clinical tasks in sinusand otology imaging. As illustrated in Fig. 9, the three si-nus protocols were each assessed as generally acceptablewith respect to high-contrast visualization of the frontal, eth-moid, maxillary, and sphenoid air cells, lamina papyracea,and skull base (including the carotid canals, vidian canal, and

Key S1 S2 S3

E1 E2 E3 E4

400

600

800

1000

1200

1400

1600

1800

123

4

5

6

7

8

9

11

11

12

13

12 13 13

13

14 1516

17

171819

20

21

10

10

11

12

13

14

10

10

13

10 17

13

11

12

13

10

FIG. 7. Spatial resolution (lp/cm) qualitatively assessed in line-pair images of the CTP528 module for each scan protocol. The approximate limiting resolutionis indicated by arrows, and the dotted box in the key denotes the approximate FOV of the unilateral protocols (E2-E4). The unilateral ear scan protocols (E2, E3,and E4) demonstrate the highest spatial resolution.

Medical Physics, Vol. 39, No. 8, August 2012

4940 Xu et al.: Technical assessment of a CBCT scanner for otolaryngology imaging 4940

0 0.5 1 1.5 2 2.5

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Spatial Frequency (1/mm)

|MT

F|

S1

E1

FIG. 8. Modulation transfer function (MTF) for the S1 and E1 protocols.Gaussian fits to the measured data are overlaid as a guide to the eye (not as amodel fit). The difference in the presampling MTF is attributed primarily tofiner sampling for the latter (isotropic voxel size 0.3 mm for S1 and 0.2 mmfor E1).

pituitary bulb). Based on qualitative assessment of sinusfeature visibility combined with the quantitative assessmentof dose, contrast resolution, and spatial resolution detailedabove, the S3 protocol was identified as the preferred de-fault (adult) protocol for sinus imaging. The S2 protocol wasidentified as an alternative, lower-dose protocol to be usedin situations where spatial resolution requirements were re-duced, soft-tissue requirements were increased, and/or therewas heightened sensitivity to radiation dose (e.g., pediatric orrepeat longitudinal imaging).

As illustrated in Fig. 9, axial, coronal, and sagittal viewsof the sinuses demonstrated resolution of fine anatomic de-

tails and air-bone interfaces. Figure 10 demonstrates the im-age quality for all four temporal bone protocols, which wereeach considered acceptable for visualization of mastoid aircells, semicircular canals, and cochlea. Isotropic 3D spatialresolution was identified as a significant strength, compara-ble to high-resolution temporal bone protocols in MDCT andsuitable for excellent visualization of submillimeter structuresat air-bone interfaces—for example, semicircular canal dehis-cence. Utility in soft-tissue visualization (e.g., cholesteatoma)was difficult to assess but (based on qualitative comparison inthe same phantom) was likely inferior to MDCT. Image qual-ity in the presence of metal (e.g., cochlear implant) was notassessed in the current work. Overall, the bilateral E1 pro-tocol was considered generally most useful, allowing assess-ment of both temporal bones from a single scan and facilitat-ing visualization of left-right symmetry. The unilateral proto-cols (E2, E3, and E4) require careful patient positioning andFOV placement to avoid truncation of structures of interest,and the scan orbits illustrated in Fig. 2 were considered dosi-metrically disadvantageous compared to E1 with respect todose to the anterior head. The E2 protocol was identified asa potential alternative in situations demanding increased spa-tial resolution in which a unilateral view was sufficient, butrequires careful attention on the part of the technologist to as-sure that the region of interest is within the smaller FOV.

IV. DISCUSSION AND CONCLUSIONS

A new commercially available CBCT scanner (CS 9300)for application in OHNS imaging (including maxillofacial,ENT, and otology imaging) was assessed in terms of techni-cal performance (dose, contrast resolution, and spatial resolu-tion) and applicability in a spectrum of clinical imaging tasks(qualitative assessment of anatomical visibility in the sinusesand temporal bones). The CS 9300 was found to provide

400

650

900

1150

1400

500

750

1000

1250

1500

400

700

1000

1300

1600S1 S2 S3

400

800

1200

1600

2000

400

800

1200

1600

2000

400

800

1200

1600

2000S1 S1 S1

FIG. 9. Qualitative assessment of image quality in the sinuses. (top) Axial slices from the S1, S2, and S3 protocols. (bottom) Sagittal and coronal views aboutstructures of interest in the sinuses for the S1 protocol. Note that the anthropomorphic phantom underwent a resection of ethmoid air cells and nasal septumto allow endoscopic access to the sphenoid for other experiments (endoscopic skull base surgery). Visualization of structures associated with the ethmoid aircells is evident in fine details of residual ethmoid along the lamina papyracea. The grayscale window and level were adjusted independently to compensate forvariations in voxel value scaling between various protocols.

Medical Physics, Vol. 39, No. 8, August 2012

4941 Xu et al.: Technical assessment of a CBCT scanner for otolaryngology imaging 4941

400

700

1000

1300

1600

300

500

700

900

1100

300

500

700

900

1100

800

1200

1600

2000

2400E2 E3 E4E1

600

1000

1400

1800

600

1000

1400

1800

400

800

1200

1600

400

600

800

1000

1200

400

600

800

1000

1200

E1

E1 E1

E2 E2

FIG. 10. Qualitative assessment of image quality in the temporal bones. (top row) Axial slices from the E1, E2, E3, and E4 protocols. (center and bottom rows)Coronal views showing structures of interest in the temporal bones for the E1 and E2 protocols, including the cochlea, stapes footplates, and bone over thesuperior semicircular canal. The grayscale window and level were adjusted independently to compensate for variations in voxel value scaling between variousprotocols.

comparable or somewhat improved radiation dose charac-teristics compared to those reported for similar application-specific CBCT scanners23 as in the studies performed byLudlow et al. for other devices. While direct comparisonto conventional (whole-body) MDCT systems is not withinthe scope of the current work, general considerations of spa-tial resolution, radiation dose, cost, and site logistics (hospi-tal versus office-based) suggest relative merits of specializedCBCT and MDCT systems. Qualitatively, the results suggestthat CBCT offers reduced radiation dose and comparable orsomewhat superior spatial resolution in comparison to com-mon MDCT protocols, but soft-tissue contrast resolution isreduced. Cost and site requirements are likely advantageousto the simpler, application-specific CBCT systems.

Image quality was judged satisfactory for high-contrast vi-sualization tasks in sinus and temporal bone imaging, withisotropic spatial resolution identified as a significant strength.Soft-tissue contrast resolution was somewhat limited in thecurrent implementation. The scanner included a number ofapplication-specific scan protocols, with the S2 and E1 pro-tocols judged to be most generally useful in terms of clini-cal utility, satisfaction of image quality requirements, and re-spectful of the desire for low-dose acquisition techniques.

A number of observations and recommendations can beappreciated in considering the dose maps of Fig. 2. The firstis the obvious challenge posed by short-scan orbits to con-ventional dosimetry standards: in addition to the limitationsin CTDI noted by Dixon et al.33 is the fact that the centraldose, Do, alone does not differentiate between protocols thatbetter spare the anterior of the head (e.g., S2 versus E4, eachwith Do ∼3.3 mGy, but differing in anterior dose by a fac-tor of 4). A simple variation on the “weighted” Dw was sug-gested to incorporate the average peripheral dose as a some-

what more useful approximation for short-scan techniques.As currently implemented by the manufacturer, the unilateralprotocols (E2, E3, and E4) shift the scan orbit off-center andlaterally. For imaging of the right ear (which was the casefor all unilateral cases in this work), the scanner shifts later-ally to the left, thereby depositing the highest dose outside theFOV. We hypothesized that an improvement in quantum noiseand sampling characteristics would be achieved by shifting in-stead to the right (not the left), placing the FOV on the regionreceiving a higher dose (reduced quantum noise) with higherdensity of backprojected rays (for a short-scan orbit). We alsonoted that all of the short-scan orbits (specifically, E2, E3, andE4) could be better constrained as in S2, S3, and E1 such thatthe x-ray tube passes posterior to the head in order to spare an-terior dose. These recommendations were relayed to the man-ufacturer to be considered in future implementations. Therewas no capability for mA modulation in the current imple-mentation, although this might allow further dose reduction ifproperly implemented.

The technical assessment performed above was performedpreliminary to a patient trial to be conducted at our institution.All of the results reported above were based on a second tech-nical assessment – the first assessment highlighting a numberof potential improvements that were constructively incorpo-rated by the manufacturer. The main recommendations high-lighted in the first assessment were: (1) a reduction in kVp foreach protocol by 5–10 kVp to the values shown in Table II;(2) a reduction in mAs by ∼10%–30% to the values shownin Table II; and (3) an adjustment of the short-scan orbitssuch that the x-ray tube traverses the posterior of the headin the short-scan orbits (as shown in Fig. 2 for S2, S3, andE1) and not the lateral aspect of the head (which imparteda significantly higher anterior dose). The first and second

Medical Physics, Vol. 39, No. 8, August 2012

4942 Xu et al.: Technical assessment of a CBCT scanner for otolaryngology imaging 4942

recommendations were based on quantitative assessment ofdose and SDNR, recognizing that the system was primarilyproviding visualization of high-contrast structures (moresothan soft-tissue) and that task performance could be main-tained even at the reduced dose levels. The third recommenda-tion (posterior short-scan orbits) was adopted for the S2, S3,and E1 protocols as shown in Fig. 2, and incorporation in allcases is being considered by the manufacturer in future im-plementations. These modifications preliminary to a clinicaltrial demonstrate the value of rigorous technical assessmentin improving the translation of technologies to clinical use.

ACKNOWLEDGMENTS

The authors extend particular gratitude to Dr. JonathanLewin and Dr. John A. Carrino (Russell H. Morgan Depart-ment of Radiology) for logistical support and valuable con-versations on image quality and deployment of dedicatedimaging systems in nonradiology medical specialties. Dr. Kr-ishnamoorthy Subramanyan (Carestream Health, Rochester,NY) is gratefully acknowledged for expertise, technical assis-tance, and valuable discussion regarding operation and per-formance of the scanner. The work was supported in partby academic-industry collaboration with Carestream Health(Rochester, NY) preliminary to a clinical trial of the scannerin the Department of Otolaryngology-Head and Neck Surgeryat Johns Hopkins University.

a)Author to whom correspondence should be addressed. Electronic mail:jeff.siewerdsen@jhu.edu; Telephone: 443-287-6269; Fax: 410-955-9826.

1L. A. Feldkamp, L. C. Davis, and J. W. Kress, “Practical cone-beam algo-rithm,” J. Opt. Soc. Am. A 1(6), 612–619 (1984).

2A. K. Hara et al., “Iterative reconstruction technique for reducing bodyradiation dose at CT: feasibility study,” Am. J. Roentgenol. 193(3), 764–771 (2009).

3J. K. Mah, R. A. Danforth, A. Bumann, and D. Hatcher, “Radiation ab-sorbed in maxillofacial imaging with a new dental computed tomographydevice,” Oral Surg. Oral Med. Oral Pathol. Oral Radiol. Endod. 96(4), 508–513 (2003).

4P. Mozzo, C. Procacci, A. Tacconi, P. Tinazzi Martini, and I. A. BergamoAndreis, “A new volumetric CT machine for dental imaging based on thecone-beam technique: preliminary results,” Eur. Radiol. 8(9), 1558–1564(1998).

5S. C. White and M. J. Pharoah, “The evolution and application of dentalmaxillofacial imaging modalities,” Dent. Clin. North Am. 52(4), 689–705(2008).

6K. Tsiklakis et al., “Dose reduction in maxillofacial imaging using lowdose Cone Beam CT,” Eur J. Radiol. 56(3), 413–417 (2005).

7J. Davies, B. Johnson, and N. Drage, “Effective doses from cone beamCT investigation of the jaws,” Dentomaxillofac Radiol. 41(1), 30–36(2012).

8M. Lamichane, N. K. Anderson, P. H. Rigali, E. B. Seldin, and L. A. Will,“Accuracy of reconstructed images from cone-beam computed tomographyscans,” Am. J. Orthod. Dentofacial Orthop. 136(2), 156–157 (2009).

9F. M. Jadu, M. L. Hill, M. J. Yaffe, and E. W. Lam, “Optimization of expo-sure parameters for cone beam computed tomography sialography,” Den-tomaxillofac Radiol. 40(6), 362–368 (2011).

10K. Araki et al., “Characteristics of a newly developed dentomaxillo-facial x-ray cone beam CT scanner (CB MercuRay): System configu-ration and physical properties,” Dentomaxillofac Radiol. 33(1), 51–59(2004).

11W. De Vos, J. Casselman, and G. R. J. Swennen, “Cone-beam computer-ized tomography (CBCT) imaging of the oral and maxillofacial region: a

systematic review of the literature,” Int. J. Oral Maxillofac Surg. 38(6),609–625 (2009).

12J. A. Roberts, N. A. Drage, J. Davies, and D. W. Thomas, “Effective dosefrom cone beam CT examinations in dentistry,” Br. J. Radiol. 82(973), 35–35 (2009).

13D. Schulze, M. Heiland, H. Thurmann, and G. Adam, “Radiation expo-sure during midfacial imaging using 4- and 16-slice computed tomography,cone beam computed tomography systems and conventional radiography,”Dentomaxillofac Radiol. 33(2), 83–86 (2004).

14H. D. Curtin, “Superior semicircular canal dehiscence syndrome and mul-tidetector row CT,” Radiology 226(2), 312–314 (2003).

15R. T. Penninger, T. S. Tavassolie, and J. P. Carey, “Cone-beam volumetrictomography for applications in the temporal bone,” Otol Neurotol. 32(3),453–460 (2011).

16M. Dahmani-Causse et al., “Morphologic examination of the temporalbone by cone beam computed tomography: Comparison with multislice he-lical computed tomography,” Eur Ann. Otorhinolaryngol. Head Neck Surg.128(5), 230–235 (2011).

17W. T. Yang et al., “Dedicated cone-beam breast CT: Feasibility study withsurgical mastectomy specimens,” AJR. Am. J. Roentgenol. 189(6), 1312–1315 (2007).

18J. De Cock et al., “Cone-beam computed tomography: a new low dose,high resolution imaging technique of the wrist, presentation of three caseswith technique,” Skeletal Radiol. 41(1), 93–96 (2012).

19I. S. Grills et al., “Image-guided radiotherapy via daily online cone-beamCT substantially reduces margin requirements for stereotactic lung radio-therapy,” Int. J. Radiat. Oncol., Biol., Phys. 70(4), 1045–1056 (2008).

20D. A. Jaffray, J. H. Siewerdsen, J. W. Wong, and A. A. Martinez, “Flat-panel cone-beam computed tomography for image-guided radiation ther-apy,” Int. J. Radiat. Oncol., Biol., Phys. 53(5), 1337–1349 (2002).

21M. A. Rafferty et al., “Investigation of C-arm cone-beam CT-guidedsurgery of the frontal recess,” Laryngoscope 115(12), 2138–2143 (2005).

22M. E. Guerrero et al., “State-of-the-art on cone beam CT imaging for pre-operative planning of implant placement,” Clin. Oral Investig. 10(1), 1–7(2006).

23J. B. Ludlow, L. E. Davies-Ludlow, S. L. Brooks, and W. B. Howerton,“Dosimetry of 3 CBCT devices for oral and maxillofacial radiology: CBMercuray, NewTom 3G and i-CAT,” Dentomaxillofac Radiol. 35(4), 219–226 (2006).

24M. Loubele et al., “Image quality vs radiation dose of four cone beamcomputed tomography scanners,” Dentomaxillofac Radiol. 37(6), 309–318(2008).

25R. L. Dixon et al., “Comprehensive Methodology for the Evaluation ofRadiation Dose in X-Ray Computed Tomography: Report of AAPM TaskGroup 111: The Future of CT Dosimetry,” AAPM Reports, AAPM TaskGroup Report No. 111, 2010.

26ICRP, “The 2007 recommendations of the International Commission onRadiological Protection. ICRP publication 103,” Ann. ICRP 37(2–4), 1–332 (2007).

27H. Watanabe, E. Honda, and T. Kurabayashi, “Modulation transfer func-tion evaluation of cone beam computed tomography for dental use with theoversampling method,” Dentomaxillofac Radiol. 39(1), 28–32 (2010).

28J. D. Silverman, N. S. Paul, and J. H. Siewerdsen, “Investigation of lungnodule detectability in low-dose 320-slice computed tomography,” Med.Phys. 36(5), 1700–1710 (2009).

29I. Pantos et al., “Adult patient radiation doses from noncardiac CT exam-inations: A review of published results,” Br. J. Radiol. 84(1000), 293–303(2011).

30M. J. Daly, J. H. Siewerdsen, D. J. Moseley, D. A. Jaffray, and J. C. Irish,“Intraoperative cone-beam CT for guidance of head and neck surgery: as-sessment of dose and image quality using a C-arm prototype,” Med. Phys.33, 3767–3767 (2006).

31D. Letourneau et al., “Online planning and delivery technique for radiother-apy of spinal metastases using cone-beam CT: Image quality and systemperformance,” Int. J. Radiat. Oncol., Biol., Phys. 67(4), 1229–1237 (2007).

32Y. Nomura, H. Watanabe, E. Honda, and T. Kurabayashi, “Reliability ofvoxel values from cone-beam computed tomography for dental use in eval-uating bone mineral density,” Clin. Oral Implants Res. 21(5), 558–562(2010).

33R. L. Dixon, “A new look at CT dose measurement: beyond CTDI,” Med.Phys. 30(6), 1272–1280 (2003).

Medical Physics, Vol. 39, No. 8, August 2012