WC 2003 August 2003

Sue Edyvean, Nicholas Keat ImPACT* (Imaging Performance Assessment of CT Scanners)LondonUK

*An MHRA Evaluation centre for the UK Department of Health

(Medicines and Healthcare productsRegulatory Agency)

www.impactscan.org

Comparison of Definitions of Geometric Efficiency in Computed Tomography Scanners

WC 2003 August 2003

Z-axis geometric efficiency in CT

• Characterises the extent of the radiation beam that is used for image creation (along the z-axis)

z-axis

WC 2003 August 2003

Geometric efficiency (g.eff.)

• Single slice – Usually little wasted dose,

as all of beam is used forimaging, except for narrow slices where sometimes post-patientcollimation is used

WC 2003 August 2003

Geometric efficiency (g.eff.)

• Multi slice– Usually more unused dose, as an

even irradiation of slices is required– Penumbral region of the beam

therefore can’t be used for imaging

WC 2003 August 2003

• IEC 60601- 2- 44 Ed. 1 1999: Original definition– “the full width at half maximum of the sensitivity profile

expressed as percentage of the full width at half maximum of the dose profile.”

• IEC 60601- 2- 44 Ed. 2 Am.1 2003: New definition – “the integral of the dose profile along the z-direction,

integrated over the range subtended by the detector elements used during acquisition, expressed as a percentage of the total integral of the dose profile in the z-direction”

• g.eff. to be displayed on scan console when less than 70%

Definitions for z-axis geometric efficiency (g.eff.)

WC 2003 August 2003

• Old definition:– Ratio of the measured imaged width to the measured

dose profile width (add imaged widths for multi-slice)

• New definition:– Integral of dose profile over range of nominal slice width

divided by integral of whole dose profile

Definitions – paraphrased

measured imaged width

dose profile width

integral ofdose profile overrange of nominal slice

nominal slice width

z-axis integral of whole dose profile

add -> total imaged width total nominal acquisition width(nominal total collimation)

WC 2003 August 2003

Purpose of this study

• Identify problems associated with old g.eff.• Compare two calculation approaches for new g.eff.• Compare values between old and new

WC 2003 August 2003

Methods of measurement – dose profiles

• Kodak X-Omat V Radiotherapy film• Scanned at iso-centre• Read out with scanning micro-densitometer• Optical density profile converted to dose profile

dose profile

0

1

2

3

4

5

6

7

8

0 5 10 15 20 25 30 35 40

distance along z-axis

dose

fwhm

WC 2003 August 2003

Methods of measurement – imaged slice width

• Angled metal plates scanned in axial mode• Width projected into image → Z-sensitivity profile

WC 2003 August 2003

Problems with old g.eff.

• Different measurement conditions (air vs phantom)• For multi slice scanners

– Many image width combinations for one beam width• Many measurements needed

– e.g. Philips Mx8000 four slice scanner

20

16

10

4

1

Beam width (mm)

2 x 104 x 520

2 x 816

1 x 102 x 54 x 2.510

1 x 42 x 24 x 14

1 x 12 x 0.51

Image width combinationsNo. x slice (mm)

Beam width (mm)

WC 2003 August 2003

Problems with old g.eff. (continued)

• Different focal spots– Potentially doubles the number of values

• Sixteen slice scanners – narrow slices– Need accurate measurement for imaged widths ~ < 1 mm– Not many existing test objects will do that

ImPACT’s ‘thin slice’ tool50 µm Titanium plates8° angle to scan plane

WC 2003 August 2003

Potential problems for calculation of new g.eff.

• Dose profile– ..‘integrated over the range subtended by detector

elements used’ ( � nominal total acquisition width)– Position of dose profile relative to detectors is not defined– Shape asymmetric � assume a reference position

dose profile

0

1

2

3

4

5

6

7

8

0 5 10 15 20 25 30 35 40

distance along z-axis

dose

WC 2003 August 2003

Two approaches for new g.eff. calculation

• Could either– Centre integration range on centre of dose profile

• Define as halfway between fwhm

– Position of range which gives maximum geometric efficiency

dose profile

0

1

2

3

4

5

6

7

8

0 5 10 15 20 25 30 35 40

distance along z-axis

dose

dose profile

0

1

2

3

4

5

6

7

8

0 5 10 15 20 25 30 35 40

distance along z-axis

dose

A. B.

WC 2003 August 2003

Two approaches for new g.eff. calculation

dose profile

0

1

2

3

4

5

6

7

8

0 5 10 15 20 25 30 35 40

distance along z-axis

dose

dose profile

0

1

2

3

4

5

6

7

8

0 5 10 15 20 25 30 35 40

distance along z-axis

dose

• Comparison of two techniques– Maximum difference = 0.04 % (ss), 0.6 % (ms)– Mean difference = 0.1% (ms)

2. Maximum g.eff.1: Centred on dose profile

A. B.

WC 2003 August 2003

Results

• Analysed data from 13 scanners (all manufacturers)– 4 single slice– 5 four slice– 1 eight slice– 3 sixteen slice

• Total of 123 different slice width combinations– Old g.eff.

• Compared new and old g.eff.

WC 2003 August 2003

Single slice - old g.eff. (imaged width / dose width)

– Most scanners have values of approximately 100%– Values higher than 100% probably due to scatter– Low for narrow slices using post patient collimation

0

20

40

60

80

100

120

0 1 2 3 4 5 6 7 8 9 10Slice width (mm)

Geo

met

ric

Eff

icie

ncy

(%)

Scanner AScanner BScanner CScanner D

70%

WC 2003 August 2003

Single slice - new g.eff. (dose in nominal width/whole profile)

– g.eff. values lower than 100% – Radiation falling outside the nominal width– Non rectangular dose profiles as widths gets thinner

0

20

40

60

80

100

0 1 2 3 4 5 6 7 8 9 10Slice width (mm)

Geo

met

ric

Eff

icie

ncy

(%)

Scanner AScanner BScanner CScanner D

70%

WC 2003 August 2003

Slice width (mm) 1 slice 2 slice 3 slice 4 slice 6 slice 12 slice

0.6 71%

1 106%

0.75 77%

1.5 77% 88%

3 88% 76%

4.5 77% 87%

5 95%

6 87%

9 77% 88%

10 96%

Multi slice - old g.eff. (imaged width / dose width)

• Unused penumbra – values less than 100%• Many values for each scanner

– e.g. Siemens Sensation 16 (small focus only)Imaged width

(mm)

WC 2003 August 2003

Slice width (mm) 1 slice 2 slice 3 slice 4 slice 6 slice 12 slice

0.6 71%

1 106%

0.75 77%

1.5 77% 88%

3 88% 76%

4.5 77% 87%

5 95%

6 87%

9 77% 88%

10 96%

Multi slice - old g.eff. (continued)

• Same beam width– Different values

Imaged width (mm)

9 mm beam

18 mm beam

WC 2003 August 2003

Multi slice - new g.eff. (dose in nominal width/whole profile)

• Values for each beam width only• Unused penumbra – values less than 100%

dose within nominal width equals total dose

dose within nominal width less than total dose

Four slice scanners only

0

20

40

60

80

100

0 2 4 6 8 101214161820222426283032Collimation (mm)

Geo

met

ric

Eff

icie

ncy

(%)

Scanner AScanner BScanner CScanner D

70%

WC 2003 August 2003

0

20

40

60

80

100

120

0 20 40 60 80 100 120Geometric efficiency (old) (%)

Geo

met

ric

Eff

icie

ncy

(new

) (%

)

Multi slice - new g.eff. versus old g.eff.

• Mean ratio = 0.99, but standard deviation = 10%• Biggest differences at narrow collimations

– total measured imaged width matches dose width, but both greater than nominal

• Some valuesnow below70% 2 x 1 mm

2 x 0.63 mm

2 x 0.63 mm

70% old

70% new

WC 2003 August 2003

Conclusions

• Old g.eff. definition– Worked well enough for single slice systems– Drawback is many slice width combinations for multi slice

• New g.eff. does not use imaged width; only the dose profile and nominal width– Two calculation techniques are comparable– Results for single slice are all generally lower

• spread of dose outside of nominal width not always identified before

– Multi-slice new g.eff. values similar to old values• except for some thin slices where both imaged slice width and dose

width are > nominal� Some values now below 70% threshold

WC 2003 August 2003

Sue Edyvean, Nicholas Keat ImPACT* (Imaging Performance assessment of CT Scanners)LondonUK

*An MHRA Evaluation centre for the UK Department of Health

(Medicines and Healthcare productsRegulatory Agency)

www.impactscan.org

Comparison of Definitions of Geometric Efficiency in Computed Tomography Scanners