abc MHRA Philips Mx8000 IDT CT scanner technical evaluation September 2004 Best choice best practice nww.medical-devices.nhs.

Transcription

1 abc September 2004 MHRA Philips Mx8000 IDT CT scanner technical evaluation Best choice best practice nww.medical-devices.nhs.uk

2 About MHRA evaluation reports. What you can expect. The Device Evaluation Service (DES) aims to provide independent and objective evaluations of medical devices available on the UK market. Specialist centres, mainly in NHS Trusts, do the evaluations under long term contract to, and in accordance with protocols approved by, the Medicines and Healthcare products Regulatory Agency (MHRA). The evaluations are usually of a unit supplied by the manufacturer. We would expect this unit to be representative of the product on the market but cannot guarantee this. Prospective purchasers should satisfy themselves with respect to any modifications that might be made to the product type after MHRA s evaluation. The reports are intended to supplement, not replace, information already available to prospective purchasers. The MHRA DES does not have access to any information held by the Agency in its capacity as the Competent Authority for the UK, apart from any information already in the public domain. The reports will contain data given by the manufacturer on the regulatory status of their devices but, apart from this, they are not an indicator of the regulatory status of a product. Occasionally, DES refers products to the regulatory arm of the MHRA for considerations of breaches of the legislation governing medical devices. DES plays no further part in any regulatory investigation that ensues and does not have advance notification of any regulatory action that may follow. How to obtain MHRA reports. To order evaluation reports, a copy of the publications catalogue or to sign up for our alert service contact: MHRA Orders Department Room 1207 Hannibal House Elephant & Castle London SE1 6TQ T: F: E: des@mhra.gsi.gov.uk Visit for a comprehensive list of publications, details of forthcoming evaluations, services and contacts. Colour reports. Full colour versions of all reports published after 2002 are available from the NHSnet at: nww.medical-devices.nhs.uk

3 Philips Mx8000 IDT CT scanner technical evaluation Assessed at Sevenoaks Diagnostic Centre, Sevenoaks, UK, July 2003 Additional tests carried out in December 2003 D Platten, N Keat, M Lewis, S Edyvean ImPACT Bence Jones Offices St George s Hospital London SW17 0QT T: F: E: impact@impactscan.org For more information on ImPACT visit Crown Copyright 2004 Apart from any fair dealing for the purposes of research or private study, or criticism, or review, as permitted under the Copyright, Designs & Patents Act, 1998, this publication may only be reproduced, stored, or transmitted in any form or by any means with the prior permission, in writing, of the Controller of Her Majesty s Stationery Office (HMSO). Information on reproduction outside these terms can be found on the HMSO website ( or hmsolicensing@cabinet-office.x.gsi.gov.uk. The MHRA is an executive agency of the Department of Health. ISBN

4 Contents Introduction...5 Philips Mx8000 IDT...5 Assessment summary...5 Specifications...6 Scanner performance clinical scans...8 Scanner performance dose and image quality...9 Head scanning...9 Body scanning...9 Scanner performance image noise...10 Variation of image noise with scan parameters...10 Inter-slice noise variation...12 Scanner performance - CT number and uniformity...13 Scanner performance - spatial resolution...14 Variation of spatial resolution with scan parameters...14 Spatial resolution and image noise...15 Limiting resolution and comparison with Philips data...16 Scanner performance - slice width characteristics...17 Imaged slice width - axial...17 Inter-slice imaged slice-thickness variation - axial...18 Imaged slice-width - helical...19 Limiting resolution along the z-axis...19 Dose profiles...20 Z-axis geometric efficiency...21 Scanner performance dose...22 CTDI 100 in air...22 CTDI 100 in Perspex head phantom...23 CTDI 100 in Perspex body phantom...23 Comparison with Philips values...23 Cone-beam reconstruction...24 Low contrast detectability...26 Low contrast detail detectability... Error! Bookmark not defined. Appendix 1: Mx8000 IDT and Brilliance CT 16 Power...27 Dose...27 Image noise...27 Imaged slice thickness - axial...28 Appendix 2: Manufacturer s comments...29 Comments from Philips...29 ImPACT response to the above comments...29 Appendix 3: Image quality assessment and Q...30 Appendix 4: ImPACT and the MHRA...31 Background...31 ImPACT...31 MHRA support to purchasers and users...31

5 Philips Mx8000 IDT Introduction The Philips Mx8000 IDT is a third generation 16-slice helical CT scanner, with a 60 kw generator, 6.5 MHU tube and a fastest gantry rotation time of 0.5 seconds. The main available collimation settings are 16 x 0.75 mm and 16 x 1.5 mm. Other settings are available, allowing smaller numbers of wider slices to be acquired, such as 8 x 3 mm, 4 x 6 mm and 2 x 12 mm. A narrow 2 x 0.6 mm collimation is also possible. The scanner s solid-state detector consists of 24 parallel rows, covering 24 mm in the z-direction at the iso-centre. The Mx8000 IDT was a development of the 4-slice Mx8000 scanner, the main difference being that the detection system has been upgraded to acquire 16-slices. The image reconstruction method used for helical scanning on the 16-slice systems is COBRA (COne Beam Reconstruction Algorithm). Since the assessment of this scanner, Philips have developed the Mx8000 scanner range, leading to the launch of the Brilliance CT range. Differences between the two ranges include a new Philips MRC x-ray tube, the design of the software userinterface, the use of optical slip-rings, and a new housing for the scanner gantry. The data presented in this report was acquired from a Mx8000 IDT model. ImPACT have tested a Brilliance CT 16 Power system in order to compare its imaging and dosimetry performance to the Mx8000 IDT. This comparison can be found in Appendix 1. Assessment summary The scanner s dose efficiency figure, Q 2, is average for both head and body scanning when compared to the sixteen-slice scanners that ImPACT have assessed. Scanners with high Q 2 values will produce images with lower measured noise for a standard patient dose and spatial resolution High contrast spatial resolution performance of the scanner is typical for a top range system. The MTF 10% figure is close to the average of the range of 16-slice scanners that ImPACT have tested, but the MTF 50% is below average. The axial inter-slice variation of image noise and slice-thickness is low, although there is a slight increase in image noise for images produced from the outer detectors. This is typical for a multi-slice scanner. There is a corresponding decrease in the slice-thickness of the outer pair of images compared to the remaining images. The helical image slice width varies very little with pitch on this scanner. A range of pitches is available, from 0.13 to 1.7. The limiting spatial resolution along the z-axis in helical scanning is below the average for the range of 16-slice scanners that ImPACT have assessed. ImPACT s assessment of low contrast detectability on the Catphan phantom showed that 0.3 % contrast details of 5 mm in diameter were visible in more than 50 % of images using a surface dose of 25 mgy. This is an improvement on the performance of the 4-slice Mx8000 scanner. 5

6 Specifications Scanner gantry Generation 3 rd Aperture (cm) 70 Maximum scan field of view (cm) 50 Nominal slice widths for axial scans (mm) 0.6, 0.75, 1.5, 3, 4.5, 6, 12 Couch Length and width (cm) 243 x 67.5 Horizontal movement range (cm) 200 Vertical movement range, out of gantry (cm) Maximum weight on couch (kg) 215 (200 to perform within stated performance) Tube and generator Generator power rating (kw) 60 Anode heat capacity (MHU) 6.5 Maximum anode cooling rate (khu / Min) 730 Guaranteed tube life (revolutions) One year unlimited revolutions Detection system Number of detector elements per row 690 Effective length of each element at isocentre (number x mm) Total effective length of detector array at isocentre (mm) Future option for more slices per rotation System start-up and calibration Total start-up time (in routine use) (mins) 5 Time to perform a full set of detector calibrations (mins) Recommended frequency for performing full sets of detector calibrations Scanning Scan times (s) Helical pitches (range and increment, quoted relative to x-ray beam width, known as pitch x ) Maximum continuous scan time (s) x 0.75; 8 x Upgrade available to all Brilliance levels 2 plus warm-up time Once every 1 to 3 weeks 0.5, 0.75, 1, 1.5 (0.42 option) , freely selectable 6

7 Operator s console Number of monitors at console 2 Control methods Mouse, keyboard Image storage Total hard disk storage capacity supplied as standard (GB) Archive options 215 MOD and CD writer (standard) Image reconstruction Time from the start of data acquisition to the appearance of the 30 th image of a series (s) for: (i) standard axial brain scan 11.3 (ii) helical abdomen scan 11.3 Simultaneous scanning and reconstruction Yes 3D reconstruction 3D reconstruction software MIPs, MinIPs, SSD, MPR (arbitrary planes), 3D volume rendering, 3D virtual endoscopy (all console and workstation) Additional facilities * Independent workstation Contrast injector Contrast media bolus tracking Standard Optional Standard Real-time CT (level 1) and CT fluoroscopy (level 2) software and hardware Hard-copy imaging device Bone mineral densitometry CT angiography Dental Radiotherapy CT simulation software Prospective ECG-triggered cardiac software Retrospective ECG-gated cardiac software Image transfer and connectivity Speed of scanner / workstation connections to local area networks (Mbits/s) DICOM services on main console (SCP and SCU) DICOM services on workstation (SCP and SCU) Optional Optional Optional Standard Optional Optional Optional Optional 100 or 1000 Storage SCU and SCP; query / retrieve SCU and SCP; print; modality worklist; performed procedure step Storage SCU and SCP; query / retrieve SCU and SCP; print * Items marked as being standard or optional may vary for different purchases. 7

8 Scanner performance clinical scans The scan settings chosen for the six clinical scans, defined in ImPACT Report MDA/98/25, were given by Philips for the Mx8000 IDT, and said to be representative of protocols in clinical use. In addition, helical protocols for head, inner ear and abdomen were used. Table 1 shows the results obtained using these settings. The quoted helical pitch, pitch x is the table increment per rotation divided by the x-ray beam width. Pitch d, which is sometimes quoted, but not used in this report, is the table increment per rotation divided by the detector acquisition width. Results in italics are mean results for the following 16-slice CT scanners: GE LightSpeed 16, Philips Mx8000 IDT, Siemens Sensation 16 and Toshiba Aquilion 16. Note that the mean z-sensitivity figures result from collimations that may differ from those used for the Mx8000 IDT. Table 1: Clinical scan settings and results Scan Scan Parameters Results kv ma Time (s) Slice thicknes (# x mm) Collimation (# x mm) Pitch x Recon FOV (mm) Recon filter CTDI vol (mgy) Z-sens (mm) Noise (%) MTF 50 (c/cm) MTF 10 (c/cm) Posterior Fossa Helical Head Standard Brain Inner Ear Helical Inner Ear Standard Abdomen Helical Abdomen Low Noise Spine Hi res spine x A (16 x 1.5) x A (16 x 1.5) x UA (16 x 1.5) x E (16 x 0.75) x E (16 x 0.75) x B (16 x 1.5) x C (16 x 1.5) x A (16 x 0.75) x C (16 x 0.75)

9 Scanner performance dose and image quality Dose efficiency is a term used to describe the quality of a scanner s images relative to the radiation dose to the patient. It can be expressed in a number of ways, and ImPACT use the Q-value, which combines measurements of noise, high contrast resolution; slice thickness and dose to produce an imaging figure of merit (see Appendix 3 for more details). The Q 2 values presented in this section are for head and body imaging. The imaging parameters used for these scans are chosen to minimise slight variations that occur for different kv, slice thicknesses, scan times and reconstruction algorithms, by using standard values where possible. These are indicated below: Tube voltage: 120 kv or 130 kv when this is the standard operating kv for the scanner. Collimation: 20 mm, or the closest available setting. Image width: 5 mm, or the closest available setting. Scan time: as recommended by the manufacturer, sub-second for body scanning and 1 s or greater for head scanning. Reconstruction algorithm: the algorithm chosen for each scanner is the one that most closely matches the average standard head and body algorithm (MTF 50% of 3.4 lp/cm, MTF 10% of 6.0 lp/cm). Reconstruction field of view: 250 mm (head) and 380 mm (body). The mas setting that would result in a CTDI vol of 50 mgy for head and 15 mgy for body scanning is listed. Z-sensitivity, image noise at 50 or 15 mgy and MTF values are also shown. Mean Q 2 values are for the following sixteen-slice CT scanners: GE LightSpeed 16, Philips Mx8000 IDT, Siemens Sensation 16 and Toshiba Aquilion 16. Head scanning Table 2: Q 2 value for head scanning Scanner Filter mas for 50mGy z-sens (mm) Noise (%) MTF 50 (c/cm) MTF 10 (c/cm) Philips Mx8000 IDT A Mean Body scanning Table 3: Q 2 value for body scanning Scanner Filter mas for 15 mgy z-sens (mm) Noise (%) MTF 50 (c/cm) MTF 10 (c/cm) Philips Mx8000 IDT B Mean Q 2 Q 2 9

10 Scanner performance image noise Variation of image noise with scan parameters Table 4 shows the variation of image noise for the ImPACT noise phantoms for a range of scan parameters. The standard parameters are 120 kv, 200 ma, 1 s scan time, 12 mm collimation, 4 x 3 mm slices, UA filter in standard mode. Table 4: Variation of image noise with scan parameters Parameter Setting Relative Noise ma Scan Time (s) Collimation (mm) (no. slices x slice width) Adjusted relative noise (4 x 0.75) (2 x 1.5) (1 x 3) (4 x 1.5) (2 x 6) (1 x 12) (16 x 0.75) (8 x 1.5) (4 x 3) (16 x 1.5) (8 x 3) (4 x 6) (2 x 12) Each noise value in this section is the average noise from a minimum of 10 images. The Adjusted relative noise column gives the relative noise value after adjustment for the theoretical inverse square relationship of image noise with ma, scan time and nominal slice-width. 10

11 Scanner performance image noise Table 5 shows the variation of image noise with reconstruction filter (convolution kernel) in both head and body phantoms. Table 5: Variation of image noise with reconstruction filter Parameter Setting Relative noise Recon. Filter (Head, standard resolution) Recon. Filter (Head, high resolution) Recon. Filter (Body, standard resolution) A 1.41 B 1.97 C 2.85 D 8.59 EB 1.61 EC 1.87 UA 1.00 UB 1.29 UC 1.66 A 1.96 B 3.60 C 8.42 D 9.83 E UA 3.66 UB 4.52 UC 5.63 A 0.48 B 0.68 C 1.00 D 2.51 EC

12 Inter-slice noise variation Scanner performance image noise Table 6 and Table 7 show the variation of image noise with detector row in a multislice axial acquisition. Parameters are 120 kv, 200 ma, 1 s scan time, UA filter. Noise values in these tables are the average of 10 images. Table 6: Variation of noise with detector row (16 x 0.75 mm) Detector row Relative noise Mean 1.00 Table 7: Variation of noise with detector row (16 x 1.5 mm) Detector row Relative noise Mean

13 Scanner performance - CT number and uniformity CT number accuracy and uniformity was assessed in ImPACT s head (18.5 cm diameter) and body (34 cm diameter) phantoms. The head phantom has a bone equivalent shell to mimic a patient s skull. Regions of interest were placed at the centre of the phantom, and 1 cm in from the inside of the periphery of the phantoms at positions corresponding to north, east, south and west compass points. ImPACT expects CT number uniformity to be within 4 HU and 10 HU for head and body size phantoms, respectively. Acquisition parameters were 120 kv, 200 mas, 4 x 3 mm image slice and UA algorithm for the head measurements. Body measurements were made at 120 kv, 200 mas, 4 x 6 mm image slice and B algorithm. Table 8: CT number accuracy and uniformity for head phantom Position CT number (HU) Difference from centre (HU) Centre N E S W Table 9: CT number accuracy and uniformity for body phantom Position CT number (HU) Difference from centre (HU) Centre N E S W

14 Scanner performance - spatial resolution Variation of spatial resolution with scan parameters Table 10 shows the variation of spatial resolution with scan parameters in terms of MTF 50% and MTF 10%. These correspond to the 50% and 10% modulation transfer function values, respectively. The unit of measurement is line pairs per cm (lp/cm). All scans used 2 x 3 mm image slices. Standard resolution head: 120 kv, 400 ma, 1 s scan time, 250 mm FOV (large focus). High and ultra-high resolution head: 120 kv, 330 ma, 1 s scan time, 250 mm FOV (small focus). Body settings: 120 kv, 267 ma, 0.75 s scan time, 380 mm FOV, standard resolution. Table 10: Variation of spatial resolution with scan parameters Parameter Setting MTF 50 (lp/cm) MTF 10 (lp/cm) Recon. filter (head, large focus, standard resolution) Recon. filter (head, small focus, high resolution) Recon. filter (head, small focus, ultra high resolution) Recon. filter (body, standard resolution) Scan type A B C D EB EC UA UB UC A B C D E UA UB UC A B C D E A B C D EC Axial Helical

15 Scanner performance spatial resolution Spatial resolution and image noise Figure 1: Image noise and spatial resolution for head scanning (CTDI w 50 mgy, 12 mm collimation, 3 mm slice with noise corrected to a nominal 5 mm slice) Image noise (%) A UA B UB UC E C D Average (MTF50 & MTF10) (lp/cm) Figure 2: Image noise and spatial resolution for body scanning (CTDI w 15 mgy, 24 mm collimation, 6 mm slice with noise corrected to a nominal 5 mm slice) Image noise (%) A B C D EC Average (MTF50 & MTF10) (lp/cm) 15

16 Scanner performance spatial resolution Limiting resolution and comparison with Philips data Table 11 shows the limiting resolution as measured by ImPACT on the Mx8000 IDT, and quoted by Philips in the product data sheet. Table 11: Comparison of ImPACT and Philips limiting resolution values Philips (lp/cm) ImPACT (lp/cm) MTF MTF For comparison, the mean limiting resolution values for a range of sixteen-slice scanners that ImPACT have assessed are MTF 50% of 10.6 lp/cm and MTF 10% of 17.9 lp/cm. This data includes the GE LightSpeed 16, Philips Mx8000 IDT, Siemens Sensation 16 and Toshiba Aquilion

17 Scanner performance - slice width characteristics Imaged slice width - axial Slice widths of 2.5 mm and above were measured using inclined aluminium plates in a water-filled phantom. Narrower slices were measured using the ImPACT thin slice test tool. This uses thinner plates at a smaller angle for improved accuracy. Table 12: Imaged slice-thickness (small focus) Nominal slice width (mm) (no. slices x slice width) Measured slice (mm) Ratio (measured:nominal) 0.75 (4 x 0.75) (16 x 0.75) (4 x 0.75) (16 x 0.75) (4 x 1.5) (16 x 1.5) (16 x 0.75) (16 x 1.5) (16 x 1.5) Table 13: Imaged slice-thickness (large focus) Nominal slice width (mm) (no. slices x slice width) Measured slice (mm) Ratio (measured:nominal) 0.75 (4 x 0.75) (16 x 0.75) (4 x 0.75) (16 x 0.75) (4 x 1.5) (16 x 1.5) (4 x 0.75) (16 x 0.75) (4 x 1.5) (16 x 1.5) (4 x 3) (4 x 1.5) (16 x 1.5) (4 x 3) (8 x 3) (4 x 3) (8 x 3)

18 Scanner performance slice width characteristics Inter-slice imaged slice-thickness variation - axial Table 14 shows the measured slice-thickness at the isocentre for each of the 16 x 0.75 mm slices at small and large focal spot. Table 15 shows the same information for 16 x 1.5 mm collimation. Table 14: Inter-slice imaged slice-thickness variation (16 x 0.75 mm, small focus) Detector group Z-sensitivity (mm) Variation from mean (%) Mean Table 15: Inter-slice imaged slice-thickness variation (16 x 1.5 mm, large focus) Detector group Z-sensitivity (mm) Variation from mean (%) Mean

19 Scanner performance slice width characteristics Imaged slice-width - helical These results were measured using a 6 mm diameter, 0.05 mm thick tungsten disc. Table 16: Imaged slice-width - helical Collimation (mm) (no. slices x slice width) Pitch x Nominal image thickness (mm) Z-sensitivity (FWHM) (mm) Ratio (z-sensitivity:nominal) 16 x x x x x x x x Limiting resolution along the z-axis The spatial resolution along the z-axis was measured using the data from the helical slice-width measurements. The limiting resolution is achieved using 16 x 0.75 mm collimation, pitch x of 0.35 and a nominal image thickness of 0.8 mm. The MTF 50% and MTF 10% values can be compared to the average for four 16-slice scanners of 5.2 and 10.4 lp/cm respectively. Table 17: Limiting resolution along the z-axis Z-sensitivity (mm) MTF50 (lp/cm) MTF10 (lp/cm)

20 Scanner performance slice width characteristics Dose profiles Radiotherapy verification film was used to measure the dose profiles. The film was placed in air at the isocentre and exposed with a single axial tube rotation for each setting. Values for both small and large focal spot are given. The dose profile width is characterised by the full width at half maximum (FWHM) of the dose profile along the z-axis. A scanning microdensitometer was used to read the film. Table 18: Irradiated slice-thickness (small focus) Collimation (mm) (no. slices x slice width) Irradiated FWHM (mm) Ratio (irradiated:nominal) 3 (4 x 0.75) (4 x 1.5) (16 x 0.75) (4 x 3) (16 x 1.5) (8 x 3) Table 19: Irradiated slice-thickness (large focus) Collimation (mm) (no. slices x slice width) Irradiated FWHM (mm) Ratio (irradiated:nominal) 3 (4 x 0.75) (4 x 1.5) (16 x 0.75) (4 x 3) (16 x 1.5) (8 x 3)

21 Scanner performance slice width characteristics Z-axis geometric efficiency Geometric efficiency is a measure of x-ray dose utilisation along the z-axis. ImPACT now uses the geometric efficiency figure as specified by the IEC CT safety standard, Ed.2 (2001) Amendment 1 (2003). This defines geometric efficiency as the ratio of the integral of the dose profile falling within the active detector width to the integral of the dose profile along its total length. Table 20: Z-axis geometric efficiency Collimation (mm) (no. slices x slice width) Focal spot Geometric efficency (%) 3 (4 x 0.75) S 62 6 (4 x 1.5) S (16 x 0.75) S (4 x 3) S (16 x 1.5) S (8 x 3) S 93 3 (4 x 0.75) L 52 6 (4 x 1.5) L (16 x 0.75) L (4 x 3) L (16 x 1.5) L (8 x 3) L 89 Figure 3: Z-axis geometric efficiency 100 Geometric Efficiency (%) Small focus Large focus Collimation (mm) 21

22 CTDI 100 in air Scanner performance dose Standard parameters are 120 kv, 200 mas, 12 mm collimation, 1 s scan time, large focal spot. Table 21: CTDI in air kv CTDI 100 (head) CTDI 100 (body) (mgy/100mas) (mgy/100mas) Table 22: Variation of CTDI in air with scan parameters Parameter Setting Relative CTDI 100 Focal Spot Total collimation (mm) (# x mm) Scan Time (s) Helical scanning (adjusted for pitch) Large 1.00 Small (4 x 0.75) (4 x 1.5) (4 x 3 ) (16 x 0.75) (16 x 1.5) ( 8 x 3) Axial 1.00 Helical

23 Scanner performance - dose CTDI 100 in Perspex head phantom For 200 ma, 1 s scan time, 12 mm collimation, large focus, head mode. Table 23: CTDI in Perspex head phantom kv CTDI Centre CTDI Periphery CTDI W (mgy/100mas) (mgy/100mas) (mgy/100mas) CTDI 100 in Perspex body phantom For 200 ma, 1 s scan time, 12 mm collimation, large focus, body mode. Table 24: CTDI in Perspex body phantom kv CTDI Centre CTDI Periphery CTDI W (mgy/100mas) (mgy/100mas) (mgy/100mas) Comparison with Philips values Philips quoted CTDI values are for 24 mm collimation. ImPACT s measurements are for 12 mm collimation. ImPACT s figures in the table below have been corrected to take account of this. Table 25: Comparison between ImPACT and Philips CTDI values Phantom, position ImPACT CTDI 100 Philips CTDI 100 Ratio (mgy/100mas) (mgy/100mas) ImPACT:Philips Head, Centre Head, Periphery Body, Centre Body Periphery

24 Cone-beam reconstruction In recent years the number of slices acquired per rotation has increased from one to sixteen and above. As the number of simultaneous slices increase the x-ray beam becomes more divergent along the z-axis, making the volume less planar and more like a cone at either end. New methods of image reconstruction have been developed to avoid artefacts that would otherwise be introduced by the cone-beam effect. The Mx8000 IDT and Brilliance CT 16 scanners use COBRA, the Philips cone-beam reconstruction algorithm, to reconstruct images when scanning in helical mode. In addition to COBRA, there is an optional filter, SP, that can be switched on at the user s discretion to further suppress artefact. ImPACT imaged a 10 mm diameter Teflon rod angled at 60 to the z-axis, immersed in a water bath. The rod was scanned helically with the centre of the scanned volume at the isocentre, using a range of pitches. The scans were repeated for the SP filter on and off. Scanning parameters were 120 kv, 16 x 0.75 mm collimation, 2 mm nominal slice-thickness, and B reconstruction filter. A pitch of was used for the first pair of images, and a pitch of 1.5 for the second. For low pitches there is little artefact seen with the SP filter off (Figure 4a). The image with the SP filter on is very similar (Figure 4b). For larger pitches there is some streaking seen in images acquired with the SP filter off (Figure 5a). This artefact is reduced by switching the SP filter on (Figure 5b). 24

25 Scanner performance cone-beam reconstruction Figure 4a: Rod, pitch 0.667, SP off Figure 4b: Rod, pitch 0.667, SP on Figure 5a: Rod, pitch 1.5, SP off Figure 5b: Rod, pitch 1.5, SP on 25

26 Low contrast detectability Exposure parameters given by Philips were 120 kv, 330 ma, 0.75 s scan time, 10 mm slice and EB algorithm leading to a dose of 27 mgy. It is not specified as to whether this is surface dose to the Catphan, or CTDI in the standard Perspex phantom. The total beam collimation is also not specified. ImPACT standard parameters are 120 kv, 25 mgy surface dose and 2 x 10 mm slices. As this collimation is not available for the Mx8000 IDT a 1 x 12 mm slice was acquired with 120 kv and 173 mas to give a surface dose of 20.8 mgy, which would give noise equivalent to a 10 mm slice at 25 mgy dose. The EB algorithm was used. For ImPACT, four observers read 20 images presented randomly along with 60 images from other 16-slice systems. A series of low contrast details from 15 to 2 mm in diameter were scored for visibility. ImPACT s criteria for the smallest visible detail listed is the smallest detail size that was visible in at least 50% of the images. Table 26: ImPACT Catphan low contrast detail Smallest visible detail (mm) Nominal contrast (%) Surface dose (mgy) ImPACT % 25 Table 27: Philips Catphan low contrast detail Smallest visible detail (mm) Nominal contrast (%) Perimeter dose (mgy) Philips % 27 26

27 Appendix 1: Mx8000 IDT and Brilliance CT 16 Power Since the assessment of the Mx8000 IDT by ImPACT, Philips have introduced the Brilliance CT range of scanners. This section compares the dose and imaging performance of the Brilliance CT 16 Power, which has replaced the Mx8000 IDT. It shows that the image quality and radiation dose characteristics of the Brilliance CT 16 Power matches that of the Mx8000 IDT. Differences between the two systems, such as the new software user-interface, are not reflected in these tests. Dose Values quoted in the tables are relative values, given as the ratio of Mx8000 IDT result to Brilliance CT 16 Power result. Scan settings were 12 mm beam collimation (16 x 0.75 mm), large focus. Head kv Relative CTDI 100, Relative CTDI 100, Relative CTDI 100, air centre periphery Relative CTDI w Body kv Relative CTDI 100, Relative CTDI 100, Relative CTDI 100, air centre periphery Relative CTDI w Image noise Values quoted in the tables are relative values, given as the ratio of Mx8000 IDT result to Brilliance CT 16 Power result. Head kv ma Time Collimation Slice Recon FOV Focal Relative Filter (s) (mm) (mm) (mm) spot noise Standard resolution EB 250 L 1.01 High resolution A 250 S 1.01 Body Time Collimation Slice Recon FOV Focal Relative kv ma Filter (s) (mm) (mm) (mm) spot noise A 380 L

28 Appendix 1: Mx8000 IDT and Brilliance CT 16 Power Imaged slice thickness - axial Collimation (# Brilliance CT 16 Mx8000 IDT Relative Slice (mm) Focus x mm) Power (mm) (mm) thickness 16 x S x L x L x S Spatial resolution Standard resolution mode kv ma Time Collimation Slice Recon FOV Filter (s) (mm) (mm) (mm) MTF50 MTF10 Brilliance CT 16 Power EB Mx8000 IDT EB Relative MTF: Ultra high resolution mode kv ma Time Collimation Slice Recon FOV Filter (s) (mm) (mm) (mm) MTF50 MTF10 Brilliance CT 16 Power E Mx8000 IDT E s Relative MTF:

29 Appendix 2: Manufacturer s comments Comments from Philips Philips Medical Systems would like to thank ImPACT for the hard work put into both the assessment and subsequent production of this report on the Philips Mx8000 IDT. However, it should be noted that as of the RSNA 2003 the Mx8000 system has been superseded by our new product line the Brilliance CT range. This range has configurations ranging from 6 to 40 slices per configuration and is substantially different to the Mx8000 Range. These systems are the result of Philips major investments in CT over recent years and incorporate technology unique to Philips. The Brilliance CT has an air-cooled gantry with an optical slip ring for fast data transfer. This allows the RapidView reconstructor to operate at rate of up to 40 images/second reconstruction. All systems incorporate the legendary MRC X-ray tube that eliminates tube cooling delays and allows high demand examinations to be performed continuously. In addition to the changes to gantry, tube, table, slip ring and cooling method the Brilliance CT range has completely different software including reconstruction algorithms, making results obtained from the IDT not comparable to the new range. Some of the features available on the Brilliance range are however available as upgrades for the Mx8000 such as a 20 ips RapidView reconstructor. Cardiac capabilities are also available on all systems with temporal resolution as fast as 53ms using rate responsive protocols to produce good images even on patients with variable heart rates. Since the measurements documented within this report there have been substantial updates to the Mx8000IDT systems in both hardware and software bringing the systems to a similar performance level as the latest Brilliance CT systems. The changes incorporated over the past 18 months have resulted in improvements to image quality, workflow and system performance generally. Many of the results listed within this document have since been superseded since the original testing. The tests and results contained within this report were performed on an Mx8000 IDT 16 System with Release 2.5 software. An updated version R2.5.5 is now being installed on all systems, with a major new software Release R 3.0, planned for early in This will substantially affect some of the results obtained in this assessment. These evaluation reports focus upon the Q factor to allow comparison with other CT systems and we appreciate that there must be some method of enabling this analysis. However, as has been said before, we feel that for this figure to have relevance to the real diagnostic imaging situation, it is of paramount importance that real clinical scanning protocols are used for the evaluation. We note that whilst we have provided actual scan protocols including scan time, this may not be the case for all systems evaluated. For instance selecting a longer scan time for a particular scan may produce a better Q factor, but it would not be possible to reproduce this in a clinical situation. In addition this assessment has been performed on a machine in clinical use, not in a less realistic factory situation. ImPACT response to the above comments ImPACT would like to thank Philips for their comments on this report. We recognise the considerable advances made in the development of the Brilliance CT range. The results in Appendix 1 compare the image quality and radiation dose characteristics of the 16-slice systems from each range. ImPACT have asked Philips for example data to demonstrate the improvements in image quality mentioned in paragraph 4 above, but we have not received any data at the time of publication of this report. Whilst it is true that for some scanners, the use of long scan times can increase spatial resolution, this effect is not apparent with the standard resolution kernels used for calculation of Q. We would like to note that Q values in this report are independent of scan time. 29

30 Appendix 3: Image quality assessment and Q Image noise, scan plane spatial resolution and imaged slice width are fundamental parameters describing the amount of object information retrievable from an image, or its image quality. Radiation dose can be regarded as a 'cost' of this information. In general, it is meaningless to quote any one of these measurements without reference to the others. It is possible to incorporate dose, noise, spatial resolution and slice width into one number, using formulae derived from the relationships between image quality and dose. Figures of merit such as this can take a number of forms depending on how the various parameters are measured and quoted. ImPACT use the Q 2 value, whose formula and methods of measurement are given below. High Q 2 values result from CT scanners that produce images with lower noise at a set spatial resolution, when dose and image width are taken into account. The parameters used in Q are standard imaging performance parameters. However it should be noted that the quantification of perceived image quality is a complicated process and as such will not be fully described by the single descriptors used for each of the parameters. Comparisons between scanners are more reliable when comparing scans reconstructed with similar convolution filters. The uncertainty in quoted values of Q 2 is up to about ± 15%, with a conservative estimate of ± 10%. Q 2 is calculated as follows: Q 2 = 1 3 fav 2 σ z CTDI vol σ = image noise, expressed as a percentage for a 5 cm 2 region of interest at the centre of the field of view in the standard ImPACT water phantoms. [%] f av = spatial resolution [lp/cm], given as (MTF 50% + MTF 10% )/2, where 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). Reconstruction algorithms with standard spatial resolution values are chosen to minimise the dependency of Q 2 upon reconstruction algorithms. The reconstruction algorithm with MTF 50% and MTF 10% values as close as possible to 3.4 lp/cm and 6.0 lp/cm is used. [lp/cm] z 1 = the full width at half maximum (FWHM) of the imaged slice profile (z-sensitivity). This is measured using the inclined plates method. [cm] CTDI vol = volume weighted CT dose index. [mgy] 30

31 Background Appendix 4: ImPACT and the MHRA One of the roles of the Medicines and Healthcare products Regulatory Agency (MHRA) is to fund evaluation programmes for medical devices and equipment. The programme includes evaluation of x-ray Computed Tomography Equipment currently available on the UK market. MHRA aims to ensure that evaluation techniques keep abreast of improvements in CT imaging performance and that MHRA reports present evaluation information that is timely, useful and readily understood. ImPACT ImPACT (Imaging Performance Assessment of Computed Tomography) is the MHRA's CT evaluation facility. It is based at St George's Hospital, London, part of St George's Healthcare NHS Trust. ImPACT have developed test objects and measurement procedures suitable for inter-comparing CT scanner performance. For each CT evaluation hundreds of images are obtained from the system under test and subsequently analysed using custom written software. Dose measurements are made using ion chambers, and x- ray film is used to obtain additional x-ray dose information. MHRA support to purchasers and users The ImPACT team is available to answer any queries with regard to the details of this report, and also to offer general technical and user advice on CT purchasing, acceptance testing and quality assurance. ImPACT Bence Jones Offices St George s Hospital London SW17 0QT T: F: E: impact@impactscan.org W: MHRA contact point for general information on the CT evaluation programme: Device Evaluation Service MHRA Elephant and Castle London SE1 6TQ T: F: E: des@mhra.gsi.gov.uk W: 31