Philips ACQSIM CT - ImPACT CT Scanner Evaluation Centre

August 2002Evaluation ReportNUMBERMDA 02099Philips ACQSIM CTCT Scanner Technical EvaluationImPACT reportMDA Evaluation ReportMDA 02099£75 (Free to the NHS)

WHAT YOU CAN EXPECT FROM MDA EVALUATION REPORTSThe Device Evaluation Service (DES) aims to provide independent and objectiveevaluations of medical devices available on the UK market. Specialist centres,mainly in NHS Trusts, do the evaluations under long-term contract to, and inaccordance with protocols approved by, the MDA. The evaluations are usually of aunit supplied by the manufacturer. We would expect this unit to be representative ofthe product on the market but cannot guarantee this. Prospective purchasers shouldsatisfy themselves with respect to any modifications that might be made to theproduct type after MDA’s evaluation. The reports are intended to supplement, notreplace, information already available to prospective purchasers.© Crown Copyright 2002Apart from any fair dealing for the purposes of research or privatestudy, or criticism, or review, as permitted under the Copyright,Designs & Patents Act, 1988, this publication may only be reproduced,stored or transmitted in any form or by any means with the priorpermission, in writing, of the Controller of Her Majesty's StationeryOffice (HMSO).Enquiries concerning reproduction outside those terms should be sentto HMSO at the undermentioned address:The Copyright Unit,The Stationery Office,St Clements House,2 - 16 Colegate,NORWICH,NR3 1BQ

Table of contentsINTRODUCTION.....................................................................................................3Philips ACQSIM CT ...............................................................................................3Report structure ....................................................................................................3SUMMARY..............................................................................................................5Scanner performance ...........................................................................................5ASSESSMENT TEAM COMMENTS ......................................................................8SPECIFICATIONS ................................................................................................10SCANNER PERFORMANCE: CLINICAL SCANS...............................................12SCANNER PERFORMANCE: DOSE EFFICIENCY.............................................13SCANNER PERFORMANCE: IMAGE NOISE .....................................................14Variation of image noise with scan parameters...............................................14Further investigations of noise variation with kV and mA..............................15SCANNER PERFORMANCE: SPATIAL RESOLUTION .....................................17Variation of spatial resolution with scan parameters......................................17Limiting resolution and comparison with Philips data....................................18SCANNER PERFORMANCE: SLICE WIDTH CHARACTERISTICS...................19Imaged slice thickness .......................................................................................19Investigations of slice width artefact ................................................................19Dose profiles .......................................................................................................20Z-axis geometric efficiency ................................................................................20Helical z-sensitivity .............................................................................................21SCANNER PERFORMANCE: DOSE ...................................................................22CTDI 100 in air ........................................................................................................22CTDI 100 in Perspex head phantom .....................................................................23CTDI 100 in Perspex body phantom .....................................................................23Comparison with Philips’ CTDI 100 values ..........................................................23Half value layer....................................................................................................24SCANNER PERFORMANCE: LOW CONTRAST DETECTABILITY...................25SCANNER PERFORMANCE: RADIOTHERAPY PLANNING.............................26In-plane distance accuracy ................................................................................26Z-axis distance accuracy in helical mode.........................................................26CT number uniformity on ImPACT Head phantom ..........................................27CT number uniformity on ImPACT body phantom...........................................28Large uniformity phantom..................................................................................29CT number linearity.............................................................................................31PATIENT COUCH.................................................................................................32Couch deflection .................................................................................................32Couch incrementation accuracy........................................................................32APPENDIX 1: IMAGE QUALITY ASSESSMENT AND Q....................................34APPENDIX 2: MANUFACTURER’S COMMENTS...............................................35Philips’ response to ACQSIM CT report .............................................................35APPENDIX 3: IMPACT AND THE MDA...............................................................37Background .........................................................................................................37ImPACT ................................................................................................................37MDA support to purchasers and users.............................................................372 ImPACT Report: Philips ACQSIM CT v 1.3

Introduction• Philips ACQSIM CTThe Philips ACQSIM CT is a single-slice, fourth generation design CT scanneremploying a static ring of solid state detectors and low voltage slip-ring technologywith an on-board generator. It is supplied with helical software as standard. It sharesmany design aspects in common with the Ultra-Z CT scanner, a conventionaldiagnostic CT scanner formerly marketed by Marconi.The ACQSIM is a specialist CT scanner, designed specifically to be used as an aid inradiotherapy treatment planning. It has a gantry aperture of 85 cm, which allowspatients to be scanned in most treatment positions. The maximum scan field of viewis 60 cm. These compare to apertures and fields of view of typically 70 cm and 50cm respectively for conventional scanners. The scanner is supplied with a flatcouch top which allows for the attachment of immobilisation devices. A CTsimulationsoftware package, with the trademark name of ACQSIM, is an option withthis system.The dual focal spot x-ray tube has a heat storage capacity of 6.5 MHU and five kVpsettings for clinical use. Unlike the Ultra-Z, the ACQSIM has a single beam shapingfilter which is used for both head and body scans. There is a choice of five slicethicknesses, 2, 3, 4, 5 and 8 mm, all with pre–patient collimation. The range of fullrotation scan times is from 1 to 4 seconds.Helical scan pitches of 0.5 to 3.0 are available on the scanner and these are used inconjunction with one of four helical interpolation algorithms.• Report structureThere are two main areas of the report, specification and performance:Scanner specificationThe scanner specification is representative of the scanner at the time of productionof this report (August 2002) and is liable to change. The manufacturer has providedthe data for this section.Scanner performanceThis section presents the results of ImPACT’s imaging and dose performanceassessment of the scanner. Although manufacturers generally publish image anddose characteristics of their scanners, different measurement techniques andphantoms often make it very difficult to compare results from one scanner againstanother.The scanner performance is presented using calculated image quality parameters:image noise, scan plane resolution and z sensitivity, as well as radiation dose. Lowcontrast detectability was also assessed visually using the Catphan phantom.Numerical and subjective imaging performance assessments from phantoms can beused as a complementary process to the evaluation of clinical images.ImPACT Report: Philips ACQSIM CT v 1.3 3

IntroductionIn addition, this scanner was tested for a number of aspects of performance that areof particular importance for radiotherapy treatment planning. CT number linearityand uniformity, in-plane distance accuracy and z-axis distance accuracy were tested.Patient couch issues, such as deflection under load and incrementation accuracy arealso presented.The Philips ACQSIM CT scanner was originally assessed by ImPACT at LincolnHospital in July 2001. At this assessment a number of artefacts and unusual resultswere observed. It was also brought to ImPACT’s attention that images obtained atClatterbridge Centre for Oncology (CCO), using their large field of view uniformityphantom, produced some unexpected results. Therefore, further tests wereperformed on this scanner using CCO’s large uniformity phantom, as well as someof ImPACT’s test tools. These measurements were performed at CCO due toavailability of the phantom and of scanner time.The performance of a conventional diagnostic CT scanner is generally compared toother CT scanners designed for the same purpose. In this case, the performanceshould also be compared to that of a “Simulator CT”, a radiotherapy simulator withCT capabilities.4 ImPACT Report: Philips ACQSIM CT v 1.3

Summary• Scanner performanceThe design of the ACQSIM CT scanner is based on a conventional single-slice CTscanner, the Ultra-Z, and many features are shared with this scanner. The changes indesign are primarily to make it more suitable for obtaining scans for radiotherapytreatment planning. The advantage of the scanner in this area is its wide gantryaperture and large scan field of view, making it more versatile for setting uppatients in radiotherapy treatment positions. The scanner is installed with externallaser alignment lights to mark the transaxial, coronal and sagittal planes. Theselights are wall-mounted for improved stability. The laser light marking the sagittalplane can track along the x-axis, facilitating skin marking for localisation of thetreatment isocentre.The performance section of this report focuses mainly on the image quality anddosimetry of the scanner, therefore the scanner is necessarily compared with singleslice scanners whose primary purpose is diagnostic imaging. This considerationneeds to be borne in mind when reading this report. To place the ACQSIM scanner incontext with its specialist application, its imaging performance could also beconsidered with respect to a conventional radiotherapy simulator adapted to createCT images (‘Simulator CT’). A commercial version of this, the SLS-CT, wasformerly marketed by Philips Medical Systems. Some key features of the twosystems are compared in Table 1.Feature Philips ACQSIM CT SLS - CTGantry aperture (cm) 85 80 - 92Maximum field of view (cm) 60 50Scan time (sec) 1 - 4 55Reconstruction time (sec) 4 12 – 45Imaged slice width (mm) 2, 3, 4, 5 & 8 2, 5 & 10Limiting spatial resolution (lp/cm) 15.0 6.7Table 1: Comparison of Philips ACQSIM and SLS-CTThe performance of the ACQSIM relative to other current single slice CT scannermodels that have been assessed by ImPACT is shown in Table 3 in the ClinicalScans section of this report. Intrinsic scanner performance in ImPACT reports isjudged by the dose efficiency value, Q 2 , for head and body scans. These arepresented in the Dose Efficiency section.Dose efficiency, Q 2The intrinsic dose efficiency, Q 2 , of scanners is best compared at a similar resolutionlevel, (see ‘Image Quality Assessment and Q’ section). The scanner is thereforecompared against current, diagnostic, single slice scanners using a ‘Soft’ algorithmfor both head and body scans. The Q 2 values obtained in this way, show that in headmode the ACQSIM has a significantly lower dose efficiency than average, whilst inbody mode it has a value slightly lower than average. To achieve the same imageImPACT Report: Philips ACQSIM CT v 1.3 5

Summarynoise at an equivalent spatial resolution would require approximately twice the dosein head scanning and about 40% more dose in body scanning as on an averagesingle slice CT scanner. An alternative way of considering this, is that for the samedose and similar resolution, image noise levels would be about 50% higher on headscans and 20% higher on body scans.Performance comparison at recommended clinical scanningparametersIn clinical practice, Philips suggest the use of the ‘Standard’ algorithm for bothposterior fossa and standard abdomen scanning. Using the recommended scanparameters for the posterior fossa scan, the dose is 30% higher at a noise level thatis more than double that for the average of current single slice scanners. For thestandard abdomen scan, the dose is almost three times higher than average, at anoise level that is close to average. It should be noted however, that resolutionvalues for the Standard algorithm are also higher than average, which in partexplains the higher noise levels for a given dose.High contrast spatial resolutionSpatial resolution (represented by the 10% modulation figure, MTF 10 ) is used as theperformance indicator for high-resolution scans. For inner ear scans, the MTF 10value obtained using the ‘Bone’ algorithm recommended by Philips is slightlybelow average. If the sharper ‘Edge’ algorithm is used, the ACQSIM achieves anMTF 10 value that is closer to the average.Doses and Diagnostic Reference Levels (DRLs)The following relates to the scan parameters already discussed above relating toperformance at recommended clinical scanning parameters. Doses for most of thehead examinations are well above average, and the resultant images have noisevalues that are significantly higher than average. The recommended scan parametersresult in CTDI w values, in most of the head protocols examined, that are above theDRL (60 mGy) for a routine head, quoted in the European Guidelines for QualityCriteria in Computed Tomography. The main exception to this is the Inner Ear scan,where the dose is well below average. Doses for the body examinations are alsoabove average but at noise levels that are average or below average. Body doses areeither below, or only slightly above, the European DRL (35 mGy) for a routineabdomen scan.Slice width characteristicsThe full width at half maximum (FWHM) values of the dose profile measurementsare, with the exception of the 2 and 3 mm slice, within 10% of the nominal slicewidth. The profiles have a triangular rather than rectangular shape, probablyresulting from the relatively close positioning of the collimator assembly to theanode.Imaged slice thickness (slice sensitivity) images obtained on the ImPACT phantomcontained artefacts. On the 3 mm and 4 mm slices, the severity of the artefactsprevented the calculation of slice width in the usual way (see ‘Assessment teamcomments’). The measurements for these two slice widths were obtained on an6 ImPACT Report: Philips ACQSIM CT v 1.3

SummaryImPACT test object designed for the measurement of narrow slices. The 3 mm slicewidth value lies just outside Philips’ specification of +/- 0.5 mm.CT number uniformity, geometric accuracy, couch deflection andincrementationThe CT number uniformity was evaluated on 18.5 cm diameter head and 34 cmdiameter body phantoms. On head phantom images, an annulus of lower CT numberthan water was seen on the inside of the bone ring. This was considered to be due toan unsatisfactory beam hardening correction. The uniformity on the 34 cm diameterbody phantom was satisfactory except at very low scanning parameters.Additionally, uniformity measurements were performed on a phantom with adiameter equivalent to 58 cm of water. This exhibited gross non-uniformityartefacts. The severity of the artefact increased with decreasing mA and kV but wasnot eliminated even at the highest factors available.When scanning phantoms or patients not centred in the scan field of view, anadditional artefact is observed. This manifests itself as a partial ring or ‘halo’ ofincreased CT number at the medial edge of the object.The in-plane distance accuracy and z-axis distance accuracy in helical mode weretested and found to be accurate within measurement errors.The couch deflection under load was less than 2 mm over a distance ofapproximately 1 metre. The couch incrementation distance was found to be accurateand reproducible.ImPACT Report: Philips ACQSIM CT v 1.3 7

Assessment team commentsThe ACQSIM CT scanner is the only wide-bore system currently on the market andcan accommodate patients for radiotherapy planning in most treatment positions. Ingeneral the scanner gives a satisfactory performance.If the system is to be used as a diagnostic scanner, it should be noted that it has alow dose efficiency on head scans and will result in higher than average patientdoses and/or higher image noise. An additional drawback is the minimum slicewidth of 2 mm which results in a limited z-axis resolution.When scanning the ImPACT noise phantom, which contains a bone equivalent ring,there was a non-uniformity of CT number present within the water region. The CTnumber of the peripheral annulus was approximately 20 HU. Although theappearance of this type of artefact is that usually attributed to beam hardening,Philips say that it is due to off-focal radiation. A correction for this is to be appliedin a future software.The artefacts produced when scanning phantoms of a large, circular cross-sectionthat approach the maximum scanned field of view, render the images of littlediagnostic value if a high enough scan technique is not used. Although the geometryof the phantom is not representative of a patient, situations are likely to beencountered clinically where certain projections of the CT scan will result in thelow signal values at which this artefact is produced. This will lead to inaccurate CTnumber information, which could affect the accuracy of radiotherapy dosecalculations if a pixel-by-pixel inhomogeneity correction is used. It may also havean impact on the process of tumour delineation. When scanning large patient crosssections the scanning parameters must be selected with care in order to avoid thisartefact. Philips are investigating approaches to overcome this problem.When scanning patients, care should be taken to centre in the scan field of view asclosely as possible in order to reduce the ‘halo’ artefact described in the previoussection.Another problem encountered was in planning helical scans with a reconstructionincrement narrower than the slice width. This would occasionally cause the systemto hang-up. This can be avoided by performing the small increment reconstructionsretrospectively. It is anticipated that the problem will be corrected in the nextversion of software.ImPACT’s slice sensitivity insert suffered from unusual artefacts in the image. Theyare thought to be due to the iterative bone correction (IBC), which is automaticallyimplemented on all head scans. The artefact was present on images of the insertwhere the CT number of the imaged plates was above 250HU. This occurred forslices of 4mm and less. It is possible that the IBC is overcorrecting due to thepresence of aluminium in the ImPACT insert as it was not apparent on clinicalimages of the skull. This type of artefact has not been previously seen onevaluations of other scanners.It was observed by ImPACT that the expected relationship between image noise andmAs and kV setting was not always obtained on this scanner. It should be noted that‘adaptive filtration’ is automatically implemented at low photon flux levels. This8 ImPACT Report: Philips ACQSIM CT v 1.3

Assessment team commentssmoothing leads to less noisy images than would be expected at low techniquefactors, but with some loss of spatial resolution.It was brought to ImPACT’s attention that severe streaking artefacts were seen onsome clinical images with irregular contours, such as in the neck/shoulder region. Itis understood that since the assessment, work has been done to decrease the severityof this artefact. This should be implemented in the next version of software.ImPACT Report: Philips ACQSIM CT v 1.3 9

SpecificationsScanner gantryGeneration4 th (rotating tube-stationary detectors)Aperture diameter (cm) 85Maximum scan field of view (cm) 60Nominal slice widths for axial scans (mm) 2, 3, 4, 5, 8Couch (Q series)Length and width (cm) physicalFlat topHorizontal movement range (cm)Scannable rangeVertical movement range out of gantry (cm)239 X 42245 x 53180162475 +/- 3mmMaximum weight on couch (kg) 202Tube and generatorGenerator power rating (kW) 56Anode heat capacity (MHU) 6.5Maximum anode cooling rate (kHU/min) 830Guaranteed tube life (revolutions @ 1 sec)Detection system100,000 pro-rataNumber of detector elements 2400Detector materialCadmium tungstate crystalEffective detector size (at isocentre) (mm) 0.68Effective detector spacing (at isocentre) (mm) 0.97System start-up and calibrationTotal start-up time (in routine use)Time to perform a full set of detectorcalibrations (mins)Recommended frequency for performing fullsets of detector calibrationsScanning4 – 5 mins system boot up120 mins (performed during service)Self calibrated, checked on quarterly basisduring preventative maintenanceScan times (s) 0.6 (partial), 1, 1.5, 2, 3, 4Helical pitches (range) 0.5 – 3.0Maximum continuous scan time (s)Operator's consoleSpiral capacity limited only by tube heat capacityraw data storage & scannable table rangeNumber of monitors at console 1Control methodsMouse, keyboard10 ImPACT Report: Philips ACQSIM CT v 1.3

SpecificationsImage storageTotal hard disk storage capacity supplied asstandard (Gbytes)Archive optionsImage reconstruction268 mm tape / 2.3 Gbyte MOD (option)Time (s) from the start of data acquisition to the appearance of the 30th image of a series:(i) standard axial brain scan(iii) helical abdomen scanSimultaneous scanning and reconstruction3D reconstruction3D reconstruction softwareAdditional facilitiesIndependent workstationContrast injectorContrast media bolus trackingCT fluoroscopy software and hardwareHard-copy imaging deviceBone mineral densitometryCT angiographyDentalRadiotherapy CT simulation softwareProspective ECG-triggered cardiac softwareRetrospective ECG-gated cardiac softwareImage transfer / connectivityDICOM service classes provided by CT console(SCP and SCU)DICOM service classes provided byindependent workstation (SCP and SCU)Speed of scanner / workstation connections tolocal area networks (Mbits/s)3 sec / image3 sec / imageYesMPR, SSD, MIP, TVMIP, 4D angioOptional AcQsim 3D simulation workstation,Pinnacle / Smart Sim, Mx ViewOptionalBolus ProN/AOptional Laser ImagerOptional on WorkstationOptional on WorkstationOptional on WorkstationAcQsim or Smart Sim optionalN/AN/AStorage SCU and SCP, Query/Retrieve SCUand SCP, Print SCU, Modality worklistWIP(optional)DICOM RT, RTOG and export for AcQsimworkstation10 / 100N/A = Not availableTable2: Philips ACQSIM CT specificationsImPACT Report: Philips ACQSIM CT v 1.3 11

Scanner performance: clinical scansThe scan settings chosen for the six clinical scans, defined in ImPACT ReportMDA/98/25, were given by Philips for the ACQSIM CT, and said to berepresentative of protocols in clinical use. An additional two oncology scans havebeen included. Table 3 shows the results obtained using these settings. Mean valuesfor performance data obtained on current top of the range, standard diagnostic,single slice scanners, (report MDA 01048) are included in the table for comparisonpurposes.ScanScan parametersResultskVpmAsSlice width(mm)Scan time (s)Recon.FOV (cm)ConvolutionfilterPitch / Scanangle^CTDIw (mGy)**z-sens. (z1)(mm)Post fossa 130 375 5 1.5 250 Std Norm 72 5.0 0.68 3.9 7.2mean 56 4.9 0.31 3.2 5.9Helical p.fossa130 375 3 1.5 250 Std 1.0* 70 3.7 1.00 4.0 7.6mean 51 5.0 0.33 3.0 5.6Standardbrain130 465 8 1.5 250 Std Norm 93 8.3 0.49 3.9 7.2mean 43 8.0 + 0.27 + 3.2 5.8Inner ear(>1mm)120 150 2 1 120 Bone Norm 25 1.9 6.06 7.4 11.0mean 35 2.0 4.76 9.3 11.3Oncologyhead130 375 3 1.5 350 Soft 1.3* 56 3.8 0.62 3.5 6.3Abdomen 120 310 8 1 400 Std Over 32 8.3 1.94 3.9 6.9mean 12 8.0 + 1.84 + 3.4 6.0Helicalabdomen120 310 8 1 400 Std 1.5* 19 9.0 2.20 3.9 7.4mean 12 9.9 1.70 3.2 5.6Low noisespine130 360 4 2 120 Std Norm 38 3.9 1.36mean 25 4.0 + 1.45 + 3.8 6.6High resspine130 360 4 2 120 Bone Norm 38 3.9 3.99mean 18 4.0 + 9.20 + 7.3 9.4Oncologyabdomen130 250 3 1 500 Soft 2.0* 13 4.7 1.63 3.6 6.1^ Scan angle: Normal = 360 0 , Over = 399 0 scan angle+ Values adjusted for same nominal slice width as on AcQsim* Standard helical interpolator used** For helical scans CTDIw has been corrected for pitchNoise (%)MTF50 (c/cm)MTF10 (c/cm)Table3: Clinical scan settings and results12 ImPACT Report: Philips ACQSIM CT v 1.3

Scanner performance: dose efficiencyDose efficiency is a term used to describe the quality of a scanner's images relativeto the radiation dose to the patient. It can be expressed in a number of ways, andImPACT have normally used the 'Q-value', which combines measurements of noise,high contrast resolution, slice thickness and dose to produce an imaging figure ofmerit (see appendix for more details). A high Q value indicates a dose efficientscanner, that is, a combination of low noise and dose for a standard resolution andslice width.The Q 2 values presented in this section are for head and body imaging. The imagingparameters used for these scans are chosen to minimise slight variations that occurfor different kV, slice thicknesses, scan times and reconstruction algorithm, byusing standard values where possible:kV: 120 kV, or 130 kV.Slice thickness: 5 mm for head, 10 (or 8) mm for body.Scan time: 1.5 or 2 s for head, 1s for body.Reconstruction algorithm: the one that most closely matches the average ‘standard’head and body algorithms (MTF 50 of 3.4 c/cm, MTF 10 of 6.0 c/cm).Reconstruction field of view: 250 mm (head) and 380 mm (body).In the two tables below the Q 2 value for the Philips ACQSIM is compared with themean value for other current single slice scanners assessed (Single Slice CTScanner Comparison Report, MDA 01048).Head scanningScanner FiltermAs for z-sens Noise MTF 50 MTF 1050mGy (mm) (%) (c/cm) (c/cm)Q 2Philips AcQsim Soft 262 5.0 0.52 3.5 6.3 4.2Mean 288 4.9 0.33 3.2 6.0 6.1Table 4: Q 2 value for head scanningBody scanningScanner FiltermAs for z-sens Noise MTF 50 MTF 1015 mGy (mm) (%) (c/cm) (c/cm)Q 2Philips AcQsim Soft 158 8.3 1.88 3.4 6.2 1.6Mean 180 9.8 1.48 3.5 6.2 1.9Table 5: Q 2 value for body scanningImPACT Report: Philips ACQSIM CT v 1.3 13

Scanner performance: image noise• Variation of image noise with scan parametersParameter Setting Relative noiseRecon. filter(Head - Post fossa)Recon. filter(Body - Standardabdomen)mAScan time (s)Slice width (mm)kVSoft 0.63Standard 1.00Detail 1.62Bone 3.62Edge 6.41Lung 2.30Soft 0.66Standard 1.00Detail 1.58Bone 3.45Edge 6.06Lung 2.22100 1.54200 1.09250 1.00400 0.50Partial 1.361 1.191.5 1.002 0.883 0.744 0.64Overscan 0.982 1.343 1.004 0.865 0.778 0.6280 1.36100 1.37120 1.09130 1.00140 0.91Flat filter Filter 1 1.09100 1.00Field of view (mm)Scan type250 1.00350 1.01600 0.98Axial 1.00Helical 1.13Table 6: Variation of image noise with scan parameters on Lincoln systemTable 6 shows the variation of image noise for the ImPACT noise phantoms.Comparing absolute values of image noise is meaningless unless the phantoms areidentical; for similarly sized phantoms, however, the noise ratios should be similar,14 ImPACT Report: Philips ACQSIM CT v 1.3

Scanner performance: image noisetherefore relative noise values are quoted. Each noise value in this section wascalculated from the average noise value from 10 images.The standard parameters to which the figures are normalised are, for head: Posteriorfossa, 130 kV, 250 mA, 1.5 second scan time, 3 mm slice, Standard reconstructionfilter, no overscan, no flat filter. For body, the standard parameters are: Standardabdomen, 120 kV, 310 mA, 1 second scan time, 8 mm slice, Standardreconstruction filter, with overscan, no flat filter.The variation of noise with, mA, slice width and scan time is as predicted for theparameters used. When investigating the variation of noise with kV, very similarnoise values were obtained for 80 and 100kV, whereas a higher value wouldnormally be expected for the former. Further measurements to determine the causeof this anomalous result were performed at CCO.• Further investigations of noise variation with kV and mA% NoisekV 250 mA 100 mA80 1.24 1.02100 1.24 1.33120 0.99 1.41130 0.89 1.39140 0.82 1.27Table 7: Variation of image noise with kV and mA setting on CCO systemTable 7 shows the variation of image noise with kV for the ImPACT head phantomat two different mA settings. The scans were performed with a head protocol at ascan time of 1.5 seconds, a slice width of 3 mm and a Standard reconstructionalgorithm. Neither ‘overscan’ nor additional flat filtration was employed. Theresults are shown graphically in Figure 1. The noise does not decrease withincreasing kV and mA as expected. It appears that a smoothing filter is introducedwhen the photon flux falls below a certain level, resulting in lower than expectednoise values.1.4% Noise1.21.0250mA100mA0.850 100 150kVFigure 1: Variation of image noise with kV and mA setting on CCO systemImPACT Report: Philips ACQSIM CT v 1.3 15

Scanner performance: image noiseHigh contrast spatial resolution is not normally affected by kV and mA and isusually only performed by ImPACT at a single kV and mA setting. However, due tothe relationship seen between kV, mA and image noise, spatial resolutionmeasurements were performed with the same parameters as for the noise scans toconfirm the use of a smoothing filter. The standard ImPACT resolution phantomcontaining a high contrast edge was used and MTF curves obtained. The results ofthe MTF analysis are presented in Table 8. These show that for the same selectedreconstruction filter, there is a reduction in spatial resolution at low scan parametersettings. It is assumed that for a larger phantom the smoothing will be introduced athigher scan parameter settings.250 mA 100 mA50% MTF 10% MTF 50% MTF 10% MTFkVlp/cmlp/cm80 3.3 7.4 2.7 5.8100 4.4 8.3 3.6 7.4120 4.3 8.0 4.2 8.2130 4.4 7.7 4.4 7.8140 4.4 7.7 4.4 7.8Table 8: Variation of spatial resolution with kV and mA setting with Standardreconstruction filterThis ‘adaptive filtration’ technique is not unique to the wide-bore design as it hasbeen noted on other scanners such as the Philips Ultra-Z, on which the design of theACQSIM is based. This intentional feature reduces image noise for low techniqueprotocols, for a slight trade off in spatial resolution.16 ImPACT Report: Philips ACQSIM CT v 1.3

Scanner performance: spatial resolution• Variation of spatial resolution with scan parametersTable 9 shows the variation of spatial resolution with scan parameters, in terms ofMTF 50 and MTF 10 , the frequencies corresponding to the 50% and 10% modulationtransfer function values respectively (in line pairs per cm).The standard scan parameters to which the figures are normalised are, for headscans: Posterior fossa, 130 kV, 300 mA, 2 second scan time, 4 mm slice, standardreconstruction filter, 250 mm field of view. For body scans: Standard abdomen,120 kV, 310 mA, 1 second scan time, 8 mm collimation, standard reconstructionfilter and a field of view of 400 mm are the standard parameters used.Parameter Setting MTF 50 (c/cm) MTF 10 (c/cm)Recon. filter(Head - Postfossa)Recon. filter(Body -Standardabdomen)Field of view,Standard filter(mm)Field of view,Edge filter(mm)Focal spot(Std. filter)Scan type(Std. Filter)Soft 3.5 6.3Standard 4.0 7.1Detail 5.0 8.0Bone 6.8 9.7Edge 8.0 10.1Lung 5.7 8.3Soft 3.4 6.2Standard 3.9 6.9Detail 4.8 7.4Bone 6.8 8.8Edge 8.0 9.8Lung 5.7 8.6600 3.8 7.5350 4.0 7.1250 3.9 7.2100 3.7 7.1250 9.0 11.1120 9.1 11.280 9.0 11.1Small 4.2 7.7Large 3.9 7.2Axial 3.9 7.2Helical 4.0 7.6Table 9: Variation of spatial resolution with scan parametersImPACT Report: Philips ACQSIM CT v 1.3 17

Scanner performance: spatial resolution• Limiting resolution and comparison with Philips dataTable 10 compares the limiting resolution as measured by ImPACT and quoted byPhilips in the product data sheet. There is good agreement between the two values,assuming similar acquisition and reconstruction parameters. ImPACT values arefor the 2% MTF level, whilst Philips’ are for the 0% MTF level. ImPACT data wasacquired at 130 kV, 200 mA, 2 second scan time, 2 mm slice, Edge algorithm, 80mm FOV. Philips do not specify the acquisition parameters.Limiting resolution(lp/cm)ImPACT, 2% MTF 15.1Philips, 0% MTF 15.0Table 10: Comparison of ImPACT and Philips limiting resolution values18 ImPACT Report: Philips ACQSIM CT v 1.3

Scanner performance: slice width characteristics• Imaged slice thicknessThe standard axial ImPACT slice width insert consists of two 0.6 mm thickaluminium plates orientated at 30° to the z-axis and positioned one above the otherto form a cross. Severe artefacts were observed on the 3 and 4 mm slice widthswhen using this insert. For this reason, the FWHM measurements presented for the3 and 4 mm slice widths are those obtained using ImPACT’s ‘thin slice’ sensitivityinsert. This contains 0.05 mm titanium foil in Perspex, angled at 8º to the scanplane. The small focal spot is automatically employed for the 2 mm slicemeasurements, and the large for all others.Nominal slice (mm) Measured slice (mm)Ratio(measured:nominal)8 8.3 1.045 5.0 1.004 3.9 0.983 3.6 1.212 1.9 0.96Table11: Imaged slice thickness• Investigations of slice width artefactArtefacts similar to those on ImPACT’s slice width insert were seen on the slicewidth inserts of the Catphan and the ACQSIM QA phantom, when scanning using ahead protocol. The measurements were repeated at CCO, but in addition, scans wereperformed using a body protocol for which the iterative bone correction is notimplemented. Figure 2 shows scans performed with a 4 mm slice using (a) a headprotocol and (b) a body protocol. The artefact is absent from scans obtained usingthe body protocol.(a)Figure 2: Images of ImPACT slice width insert scanned with a 4 mm slice with (a)Head protocol and (b) Body protocol(b)ImPACT Report: Philips ACQSIM CT v 1.3 19

Scanner performance: slice width characteristicsThe artefacts seen on the slice width inserts are thought to be due to the iterativebone correction (IBC), which is automatically implemented on all head scans. Theartefact was present on images of the insert where the CT number of the imagedplates was above ~250HU. This occurred for slices of 4mm and less, where areduction in CT number of approximately 50 HU is seen on the high density side ofa water/high density material interface. It is possible that the IBC is overcorrectingdue to the presence of aluminium in the ImPACT insert, however, it has not beennoted on other CT scanners. The artefact was not apparent on images of a skullphantom or on examples of clinical head images.• Dose profilesThe dose profiles were measured using radiotherapy verification film, suspended inair, at the centre of the field of view, with a rotating tube. The small focal spot wasused for the 2 mm slice and the large for all others.Nominal slice(mm)Irradiated FWHM(mm)8 8.5 1.075 5.3 1.064 4.4 1.103 3.6 1.222 2.3 1.15Table 12: Irradiated slice thicknessRatio(irradiated:nominal)• Z-axis geometric efficiencyThe z-axis geometric efficiency is given by the ratio of imaged to irradiated slicethicknesses.Nominal slice(mm)Imaged sliceFWHM (mm)Irradiated FWHM(mm)8 8.3 8.5 97%5 5.0 5.3 94%4 3.9 4.4 90%3 3.6 3.6 99%2 1.9 2.3 83%Table 13: Geometric efficiencyGeometricefficiency20 ImPACT Report: Philips ACQSIM CT v 1.3

Scanner performance: dose• CTDI 100 in airStandard scan parameters were 120 kV, 200 mA, 1 second scan time, 8 mmcollimation, Filter 0 (no additional flat filter), ‘Normal’ (360 0 ) scan angle.kVCTDI 100(mGy/100mAs)80 14.4100 20.8120 27.7130 31.3140 35.1Table 15: CTDI in airParameter Setting Relative CTDI 100X-ray filterTotal collimation(mm)Scan time (s)mAHelical scanning0 1.001 0.618 1.005 0.954 0.953 0.962 0.98Partial 0.711 1.001.5 1.512 2.033 3.064 4.09Overscan 1.1050 0.25100 0.50200 1.00300 1.50400 1.99Axial 1.00Pitch 1 0.96Table16: Variation of CTDI with scan parameters, normalised to standard scanparameters given above22 ImPACT Report: Philips ACQSIM CT v 1.3

Scanner performance: dose• CTDI 100 in Perspex head phantomScan parameters: 200 mA, 1 second scan time, 8 mm collimation, filter 0 (noadditional flat filter), ‘Normal’ (360°) scan angle.kVCTDI Centre CTDI Periphery CTDI W(mGy/100mAs) (mGy/100mAs) (mGy/100mAs)120 15.3 18.3 17.3130 18.0 21.1 20.0Table 17: CTDI in Perspex head phantom• CTDI 100 in Perspex body phantomScan parameters200 mA, 1 second scan time, 8 mm collimation, filter 0 (noadditional flat filter), ‘Normal’ (360°) scan angle.CTDI CentrekVCTDI Periphery CTDI W(mGy/100mAs) (mGy/100mAs) (mGy/100mAs)120 4.6 11.8 9.4130 5.5 13.7 10.9Table18: CTDI in Perspex body phantom• Comparison with Philips’ CTDI 100 valuesPhantom, PositionImPACT CTDI Philips CTDI Ratio(mGy/100mAs) (mGy/100mAs) ImPACT:PhilipsHead, Centre 18.0 18.0 1.00Head, Periphery 21.1 21.5 0.98Body, Centre 5.5 5.2 1.06Body Periphery 13.7 13.0 1.05Table19: Comparison between ImPACT and Philips CTDI values at 130 kVImPACT Report: Philips ACQSIM CT v 1.3 23

Scanner performance: dose• Half value layer80kV 100kV 120kV 120 kV+filter 130kV 140kVHVL (mm Al) 3.6 4.7 5.7 8.0 6.3 6.8Table 20: Results of half value layer measurements24 ImPACT Report: Philips ACQSIM CT v 1.3

Scanner performance: Low contrast detectabilityThe system’s low contrast detail detectability was tested with a Catphan 500phantom.Catphan images were obtained with two sets of exposure parameters, firstly usingImPACT’s standard settings, and secondly with the protocol used by Philips toobtain their performance figures. The exposure parameters used to obtain astandard ImPACT surface dose of approximately 25 mGy were: 120 kV, 140 mA, 1second, 8 mm slice thickness, 250 mm reconstruction field of view, Soft algorithm.Philips’ recommended protocol for this test was: 130 kV, 400 mA, 1 second, 8 mmslice thickness, 250 mm reconstruction field of view, Soft algorithm. This resultedin a surface dose of 78 mGy.ParametersSmallest visible Nominal Surface dosedetail (mm) contrast (mGy)ImPACT standard 6 0.30% 24Philips recommended 2 0.30% 78Table 21: Catphan low contrast detailImPACT Report: Philips ACQSIM CT v 1.3 25

Scanner performance: radiotherapy planningAs the ACQSIM CT is specifically designed to be used for radiotherapy planningpurposes, a number of extra test results are included in this report.• In-plane distance accuracyThis test quantifies the geometric distortion of axial images. Leeds Test Object(M1), a Perspex disc of 38 cm diameter containing a wire grid, was used. Thephantom was set up off-centre in order to test geometric accuracy out to the edge ofthe field of view. No distortion was evident over the entire field of view of 600 mmto within the limits of accuracy of the measurement (+/- 1mm).AxisActual MeasuredDistance (mm) Distance (mm)DistortionHorizontal 320 320 0%Vertical 260 260 0%Table 22: In-plane distance accuracy• Z-axis distance accuracy in helical modeThis was tested using ball bearings located at different z-axis positions along the z-axis. A helical scan was performed, and the distance between the ball bearingsevaluated. Reconstructions were performed at 0.5 mm intervals in the region of theball bearing, which was located by finding the slice with the maximum CT number.Table 23 indicates that the distance accuracy is correct to within the limits ofaccuracy of the measurement (+/- 0.5 mm).Distance between ball Indicated distancebearings (mm)(mm)200 200.5Table 23: Z-axis distance accuracy26 ImPACT Report: Philips ACQSIM CT v 1.3

Scanner performance: radiotherapy planning• CT number uniformity on ImPACT Head phantomRegions of interest were placed at the centre and at 4 positions (‘North’, ‘East’,‘South’ and ‘West’), 2cm from the edge of ImPACT’s head phantom. This consistsof a water-filled cylinder of bone-equivalent material with Perspex ends. Theexternal diameter of the phantom is 18.5 cm, and the thickness of the boneequivalentwalls is 3 mm. The difference between the CT number of the centralregion and the CT number at the peripheral positions is shown in Table 24. TheImPACT head phantom images contained an artefact within the bone-equivalentring (see Figure 3). This was in the form of an annulus, almost 2 cm wide, ofincreased CT number within the water region. According to ImPACT’s protocol,uniformity is assessed 2 cm from the edge of the phantom, therefore themeasurements presented in Table 24 did not include this artefact and appear to besatisfactory.Position Mean CT NumberCT NumberDifferenceCentre -2.7N -2.3 0.3E -3.4 -0.8S -3.8 -1.1W -3.3 -0.6Table 24: CT number uniformity– ImPACT head phantomThe CT number of the annulus was ~12 HU above background when the phantomwas centred in the field of view. When the head phantom was scanned with a bodyprotocol i.e. with the iterative bone correction algorithm (IBC) switched off, theseverity of the artefact increased, with the annulus having a CT number of ~25 HU.Figure 3 shows the artefact with and without IBC. This artefact appears to be due tobeam hardening effects.(a)(b)Figure 3: Artefact on head phantom with (a) head protocol (b) body protocolThe Philips representatives claim that this artefact is due to off focal radiation andwork is in progress to correct this in the next version of software.ImPACT Report: Philips ACQSIM CT v 1.3 27

Scanner performance: radiotherapy planning• CT number uniformity on ImPACT body phantomThe ImPACT body phantom consists of a water-filled, Perspex cylinder with anexternal diameter of 34 cm. The Perspex walls are 1 cm thick. Uniformitymeasurements were performed on this phantom at standard abdomen parameters.The results are shown in Table 25 and Figure 4. The uniformity was within thespecifications of IPEM Report No. 77 of +/- 1.5%.Position Mean CT NumberCT NumberDifferenceCentre -1.4N 4.9 6.3E 4.3 5.7S 4.6 6.0W 4.1 5.5Table 25: CT number uniformity– ImPACT body phantomFigure 4: Uniformity on 34 cm diameter ImPACT water phantom at standardabdomen scanning parameters: 120 kV, 310 mA, 1 sec, 8 mm slice.An artefact, seen as a region of substantially decreased CT number in the centralportion of the phantom, was observed at very low scan parameter settings e.g. 120kV, 30 mA, 2 sec, 2 mm or 80 kV, 200 mA, 2 sec, 2 mm.When the phantom/patient is not centred in the field of view, a partial annulus ofhigher CT number is seen at the medial edge of the object (i.e. the portion nearestthe centre of the FOV). In Figure 5 the centre of the ImPACT body phantom is 13cm above the isocentre, but the whole phantom is still within the scanner’s 60 cmfield of view. The difference in CT number between the centre of the phantom andthe partial ‘halo’ is approximately 25 HU. This artefact was also seen on a skullphantom positioned off centre.28 ImPACT Report: Philips ACQSIM CT v 1.3

Scanner performance: radiotherapy planningFigure 5: ‘Halo’ artefact on body phantom positioned off-centreHeads scans on patients being planned for stereo-tactic treatment are performedwith the patient in an immobilisation device that raises the head into an off-centreposition. These scans contain a combination of the two non-uniformity artefactsdescribed above. The difference in CT number seen on one particular patient due tothese artefacts was ~35 HU, resulting in difficulties in delineating the treatmentvolume.• Large uniformity phantomFurther uniformity measurements were performed at CCO using their largeuniformity phantom. This consists of a water-filled, Perspex cylinder, 2 cm inlength. The phantom has an outer diameter of 57 cm, the Perspex wall is 3 cm thick.The phantom was centred in the field of view and scanned with a range ofparameters. Gross artefacts were seen, even at the highest settings available i.e. 140kV, 400 mA, 4 sec., 8 mm slice. Examples of images obtained are shown in Figure6. In Figure 6a, the central CT number is ~ -40 HU, surrounded by a region of ~ 90HU, and a peripheral region ~ 30 HU. When the kV, mA or slice width are reduced,the CT number of the central area decreases and, at the same time, its diameterincreases. In the example shown in Figure 6b, the CT number of the central area is~ -250 HU and the peripheral band is ~10 HU. The scan time did not affect theseverity of the artefact.ImPACT Report: Philips ACQSIM CT v 1.3 29

Scanner performance: radiotherapy planning(a)(b)Figure 6: Non-uniformity artefact on large diameter body phantom (a) Maximumscan technique:140 kV, 400 mA, 2 sec., 8 mm. (b) 120 kV, 100 mA, 2 sec., 8 mm.This artefact occurs because a threshold has been set by the manufacturer abovewhich the measured attenuation is not permitted to rise. As the signal is clamped atthis level the apparent attenuation appears to be less than the actual value. Underthese circumstances anomalies in CT number will occur. In service mode it ispossible to change the threshold level. The artefact then disappears but the imagebecomes far noisier.On the standard ImPACT body uniformity phantom with a diameter of 34 cm, theuniformity was satisfactory at standard scanning parameters and the artefactappeared only at very low technique factors. When scanning a water equivalentphantom whose diameter approaches the field of view of the scanner, CT numberuniformity is affected, even at the highest technique factors. On phantoms withintermediate diameters between 34 cm and 58 cm, it would be expected that theartefact would occur at settings higher than these.On large patient cross-sections, the attenuation of the x-ray beam is recorded asbeing lower than the true attenuation. This can lead to severe drops in CT number,and a loss of image contrast. Although it is unlikely that patients with cross-sectionsof 57 cm in all projections will be scanned, some projections are likely to have thislevel of attenuation with artefacts in the resulting images.30 ImPACT Report: Philips ACQSIM CT v 1.3

• CT number linearityThis was tested with an RMI 467 electron density phantomScanner performance: radiotherapy planningElectron DensityMean CT Number0.292 -693.10.438 -576.50.895 -75.90.945 -67.10.980 -24.41.020 2.91.039 28.21.050 85.11.081 208.11.099 232.41.116 97.01.142 156.61.147 133.01.285 442.91.473 770.31.707 1157.4Table 26: CT number linearity140012001000800CT Number6004002000-2000 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8-400-600-800Electron DensityFigure 7: CT number linearityImPACT Report: Philips ACQSIM CT v 1.3 31

Patient couchThe patient couch was tested for deflection under load as it is moved through thegantry and for couch incrementation accuracy.• Couch deflectionThe couch was loaded with 70 kg of water, distributed along its length. Theposition of the top of the couch relative to the centre of the field of view wasmeasured at different z-axis positions. Table 27 shows a couch deflection of justover 1 mm over a scan length of nearly 1 m.Z-axis position (mm)Couch top to centreFOV distance (mm)449 1.3-510 2.7Deflection -1.4Table 27: Patient couch deflection• Couch incrementation accuracyThis test assesses the accuracy of couch incrementation in axial mode. Figure 8shows the distribution of optical density along the length of a radiotherapyverification film, using a 2 mm slice width, and 6 mm slice increment. The resultsfrom this test are summarised in table 27 and show the couch incrementation to becorrect to within measurement accuracy limits (± 0.1 mm).1.210.80.60.40.200 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160Distance (mm)Figure 8: Couch incrementation32 ImPACT Report: Philips ACQSIM CT v 1.3

Patient couchNumber of slicesTotal distance Mean slice(mm) increment (mm)10 59.95 6.0020 120.10 6.01Table 28: Couch incrementation accuracyImPACT Report: Philips ACQSIM CT v 1.3 33

Appendix 1: Image quality assessment and QStatistical noise, spatial resolution and slice sensitivity are fundamental parametersdescribing the amount of object information retrievable from an image, or its imagequality. X-ray dose can be regarded as a 'cost' of this information. In general, it ismeaningless to quote any one of these measurements without reference to theothers. The Q-value incorporates the dose, noise, spatial resolution and slice widthinto one number. This figure is derived from a relationship between image qualityand dose received.A dose efficiency factor has a fundamental meaning, in that a dose efficient scannerwill produce good resolution at minimum dose and noise. However, it can take anumber of forms depending on how the various parameters are measured andquoted.The Q-value used in this comparison report, Q 2 , is the same one used inComparison Report 12 (MDA/00/11), which was modified from the previous valueused by ImPACT, Q 1 .Q 2 is defined as follows:Q2=fav2σ z CTDIwhere:σ = image noise, expressed as a percentage for a 5cm 2 region of interest at thecentre of the field of view in the standard ImPACT water phantoms.f av = spatial resolution, given as (MTF 50% + MTF 10% )/ 2Where MTF 50% and MTF 10% are the spatial frequencies corresponding to the 50%and 10% modulation transfer function values respectively (in line pairs per cm).z 1 = the full width at half maximum (FWHM) of the imaged slice profile (zsensitivity).This is measured using the inclined plates method for axial imaging,and using a 0.1mm thickness, 6mm diameter tungsten disc for helical scanningCTDI w = weighted CT dose index, as defined in IEC 16262The Q-factor is in part empirical and it should be used with caution. It is not anabsolute figure, as its derivation relies on assumptions of the shape of convolutionfilter used. Comparisons between scanners will be more reliable when comparingscans reconstructed with similar convolution filters. It is of most importance whenconsidering the standard scans for head or body. The uncertainty in this value is upto about +/-15%, with a conservative estimate of +/- 10%.13w34 ImPACT Report: Philips ACQSIM CT v 1.3

Appendix 2: Manufacturer’s comments• Philips’ response to ACQSIM reportThe system reviewed in this report is the ACQSIM, a CT system specificallydesigned for Radiotherapy Simulation and Planning. It is appreciated that assuch it occupies a unique position in the imaging field and also presentssome challenges to both the users and to those assessing such a system.The initial evaluation undertaken prior to clinical use revealed some unusualresults that have been investigated by both Philips and the ImPACT group.These and the subsequent tests that form the basis of this additional reviewwere performed upon systems running the first release of software on anewly launched platform. Although a relatively new system it has alreadygained acceptance and is in daily clinical use in a large number of sitesworldwide, including the UK.The results obtained in these evaluations are not disputed, but it is felt that asystem designed for such a specific use may require testing using methodsand parameters different to that used for diagnostic systems. ClatterbridgeCentre for Oncology are to be applauded for their innovation in producing aphantom that has highlighted the need for further development of thehardware and software of the AcQsim and this has been taken into accountwith Software Release 2, due mid 2002. This will address the image qualityissues raised in this report. Images from CCO reconstructed with this newsoftware will be made available to them during February 2002. Philips hasalso, together with CCO and a commercial phantom manufacturer (PhantomLabs), produced an ovoid phantom specifically to allow standard evaluationof such a large bore CT system. CCO will be undertaking further tests withthis phantom in February, when it is anticipated that many of the issuesconcerning signal loss and photon starvation will be resolved. Ourdevelopment group in Cleveland is in direct contact with CCO and weexpect a resolution of the issues raised at CCO and further highlighted inthis report in the near future.We also understand that similar evaluations of other scanners designatedas suitable for RT Simulation will now be undertaken for the purposes of theCapital Investment Programme. In addition we would wish to provideImPACT with the opportunity to evaluate the system once Release 2software has been implemented and using the Phantom Labs phantom. Wewould also welcome further discussions to establish the most appropriatemethods of assessing such systems designed specifically for radiotherapypurposes. The proposed Philips Oncology CT Users group would provide asource of information that could be useful in this process.ImPACT Report: Philips ACQSIM CT v 1.3 35

Manufacturer’s commentsWe would also like to comment further on some of the individual resultswithin this report:Large Phantom UniformityWhen absolutely circular objects are scanned at low dose levels thepresentation of photon starvation is enhanced since all source fans will haveidentical profiles, which add cumulatively during reconstruction. The reportnotes that the artifacts are much reduced when using the ImPACT phantomand are less apparent in actual clinical scans. It is possible to reduce theseeffects by altering the clamping level.Noise and spatial resolutionIn common with other systems special techniques are employed to produceacceptable image quality when photon flux falls below certain levels.Selection of factors appropriate for clinical examinations reduces this effect.Halo EffectIt is appreciated that artifacts can occur in very off-centred scans.Improvements in protocol programming and techniques minimize this inclinical use. However, further improvements to reduce artifacts due to offcentrepositioning will form part of the next software release. Changes inbeam hardening compensation will also be implemented in the next releaseof software to further enhance image quality in head scanning.Slice Ramp ImagesThe effects seen are because the iterative bone correction algorithm hasbeen calculated using a calcium spectrum to optimize image quality. This isthe reason that no artifacts are seen on clinical images or on a skullphantom, but are seen on the aluminium ramp in a water phantom.Angela NightingalePhilips Medical Systems36 ImPACT Report: Philips ACQSIM CT v 1.3

Appendix 3: ImPACT and the MDA• BackgroundOne of the roles of the Medical Devices Agency (MDA) is to fund evaluationprogrammes for medical devices and equipment. The programme includesevaluation of x-ray Computed Tomography Equipment currently available on theUK market.MDA aims to ensure that evaluation techniques keep abreast of improvements inCT imaging performance and that MDA reports present evaluation information thatis timely, useful and readily understood.• ImPACTImPACT (Imaging Performance Assessment of Computed Tomography) is theMDA's CT evaluation facility. It is based at St George's Hospital, London, part of StGeorge's Healthcare NHS Trust.ImPACT have developed test objects and measurement procedures suitable forinter-comparing CT scanner performance. For each CT evaluation hundreds ofimages are obtained from the system under test and subsequently analysed usingcustom written software. Dose measurements are made using ion chambers, and x-ray film is used to obtain additional x-ray dose information.• MDA support to purchasers and usersThe ImPACT team is available to answer any queries with regard to the details ofthis report, and also to offer general technical and user advice on CT purchasing,acceptance testing and quality assurance.ImPACTBence-Jones OfficesSt. George's HospitalLondon SW17 0QTTel: 020 8725 3366Fax: 020 8725 3969email: impact@impactscan.orgweb site: http://www.impactscan.orgMDA contact point for general information on the CT evaluation programme:CT Scanner Evaluation Co-ordinatorHannibal HouseElephant and CastleLondon SE1 6TQTel: 020 7972 8155Fax: 020 7972 8105ImPACT Report: Philips ACQSIM CT v 1.3 37

MEDICAL DEVICES AGENCYMDA Evaluation ReportsMDA evaluation reports are published by the Medical Devices Agency, an Executive Agency of the Department ofHealth. They are available free of charge to the UK National Health Service (NHS), and are for sale to commercialorganisations and other interested parties. A free catalogue of available reports can be obtained from the OrdersDepartment, or downloaded from the MDA web site:http://www.medical-devices.gov.ukOrderingSend your order to the address given below, stating the number, title and quantity of each report required. Yourreports will be despatched by second class post the following working day. If you are not a representative of theNHS, you will be invoiced separately. Non-NHS customers are reminded that it is not possible to offer refunds forreports ordered in error.EnquiriesOrders DepartmentRoom 1207Medical Devices AgencyHannibal HouseElephant and CastleLondonSE1 6TQTel: 020-7972 8181Fax: 020-7972 8105E-mail: dep@medical-devices.gov.ukGeneral publication enquiries should be directed to the Orders Department:Tel: 020-7972 8181Fax: 020-7972 8105E-mail: dep@medical-devices.gov.uk

ISBN 1 84182 600 6Smart number 36 28944 1