Imaging A Medical Devices ImPACT Agency Evaluation … · used in NRPB SR250 Then use appropriate...
Transcript of Imaging A Medical Devices ImPACT Agency Evaluation … · used in NRPB SR250 Then use appropriate...
EFOMP Workshop at ECR 2002
CT dosimetry and a data base for CTDI values
ImPACT
S. Edyvean, ImPACTSt George’s Hospital,
London
Imaging Performance
Assessment of CT Scanners
A Medical Devices Agency Evaluation Group
ImPACT EFOMP 2002 3
• Computed Tomography Dose Index (CTDI)– CTDIFDA
– CTDI100
– CTDIw
CTDIw =1/3CTDIc +2/3CTDIp
c = centre positon, p= periphery position
CT scanner dosimetry
dzD(z)T1 = CTDI100 ∫
+
−
50
50
ImPACT EFOMP 2002 4
• Monte Carlo Calculations• NRPB (1991,1993)• GSF (1991)• WinDose
CT scanner dosimetry
ImPACT EFOMP 2002 5
• NRPB SR250, 1993• 23 Organ dose data sets generated using MC calculations
→ Effective Dose• Organ dose and ED expressed relative to CTDI in air
→ nD mGy / mGy, nED mSv/ mGy
• Numerical phantom
NRPB CT organ dose datasets
NRPB R248 , 249, 250, SR250
ImPACT EFOMP 2002 6
NRPB CT organ dose datasets
• Relevant information from manufacturersfor 27 scanner models
• Condensed to 13 scanner groupsbased on scanners withsimilar dosimetry characteristics
• Different kVs23 Monte Carlo
organ dose data sets
ImPACT EFOMP 2002 7
Purpose of ImPACT survey• In 1996 80 scanner models with different names
in use in the UK• NRPB SR250 MC data sets not applicable to ~65% of
scanners in UK• increasing due to number of newer scanner models • How to estimate patient doses on these scanners ?
ImPACT EFOMP 2002 8
Three choices
1. Generate new MC datasets– no relevant expertise – difficult to acquire accurate information
on ~ 80 scanner models
ImPACT EFOMP 2002 9
Three choices2. Data set from an ‘average’ scanner
– does not address differences between models
3. Use dose distribution characteristics tomatch new scanner models to thoseused in NRPB SR250Then use appropriate MC data set– Based on assumption that old and
new scanners fell within similar ranges– method chosen
ImPACT EFOMP 2002 10
scanner z-axis
detectors
Scanner couch
x-ray tube
ion chamber
x-ray slice y-axis
tilt
Scanner Characteristics Survey• Measurement based approach
– easy to carry out on existing and future scanners• Standard protocol
– how to do the measurements– accuracy and tolerances of set up– data sheets
• 30 centres, 74 scanners• UK + Holland,France,Spain
ImPACT EFOMP 2002 11
• 80 models ~ 40 groups– at first by partial knowledge– then refined by review of measurement data
Scanner models
CTDIphantom
Perspex Phantom
Off-axis variationHVL
Aluminium
CTDIair
Detectors
Ion chamber
X-ray source
ImPACT EFOMP 2002 13
• Separated scanners into – those with NRPB SR250 MC data sets– newer models without
ImPACT survey data
ImPACT EFOMP 2002 14
Scanners with NRPB SR250 MC data sets
CTDIph
HVL
CTDIair
• Survey measurement data versus calculated ED
• Establish which combination of data gave bestcorrelation
ImPACT EFOMP 2002 15
Scanners with NRPB SR250 MC data sets
CTDIph
HVL
CTDIair
• Survey measurement data versus calculated ED
• Establish which combination of data gave bestcorrelation
ImPACT EFOMP 2002 16
Scanners with NRPB SR250 MC data sets
CTDIph
HVL
CTDIair
• Survey measurement data versus calculated ED
• Establish which combination of data gave bestcorrelation with ED
ImPACT EFOMP 2002 17
CTDI data : phantom factor
Phantom Factor = CTDIw / CTDIair
Gelijns, K. Patient Dosimetry in Diagnostic Radiology, Chest Examinations and Computed Tomography, PhD Thesis, Leiden
CTDIph
HVL
CTDIair
ImPACT EFOMP 2002 18
‘ImPACT’ factor
• Multi-variate analysis – for combination which gave best correlation
‘ImPACT’ Factor
ImF = a. (CTDIc / CTDIair)+ b. (CTDIp / CTDIair)+ constant
CTDIair CTDIc,CTDIpCTDIph
HVL
CTDIair
ImPACT EFOMP 2002 19
HVL vs normalised EDy = 7.6119x + 0.0736
R2 = 0.9131
0
2
4
6
8
10
12
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Normalised Effective Dose
HVL
(mm
)
(mSv/mGy)( ED / CTDIair )
ImPACT EFOMP 2002 20
HVL vs normalised EDy = 7.6119x + 0.0736
R2 = 0.9131
0
2
4
6
8
10
12
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Normalised Effective Dose
HVL
(mm
)
(mSv/mGy)( ED / CTDIair )
ImPACT EFOMP 2002 21
y = 0 .7 0 8 9 x + 0 .0 0 1 2R 2 = 0 .9 3 6 9
0
0 .1
0 .2
0 .3
0 .4
0 .5
0 .6
0 .7
0 .8
0 .9
1
0 .0 0 .2 0 .4 0 .6 0 .8 1 .0 1 .2 1 .4
N o rm a lis e d E ffe c tiv e D o s e
Phan
tom
Fac
tor (
CTD
I w/ C
TDI ai
r)Phantom factor vs normalised ED
( ED / CTDIair )
ImPACT EFOMP 2002 22
y = 0.977x + 0.0203R2 = 0.977
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Normalised Effective Dose (Abdomen)
ImF vs Effective Dose
( ED / CTDIair )
ImPACT EFOMP 2002 23
Correlations for data combinations
0.80.8250.85
0.8750.9
0.9250.95
0.9751
HVL CTDI CTDI HVL,CTDI
Cor
rela
tion
co-e
ffici
ent (
r^2)
headneckchestabdopelvis
(PF) (ImF)HVL CTDI CDTI HVL & CTDI
(r2
)
ImPACT EFOMP 2002 24
Newer scanner models• ImF calculated• Compare to ImFs for SR250 scanners• Find closest match• Use that MC dataset
Data Set
MC 1
MC 2
SR250 ImFScanner
xxx 0.60
yyy 0.75
New ImFScanner
Philips AV 0.62
ImPACT EFOMP 2002 30
CTDI and collimation
0.00
0.50
1.00
1.50
2.00
2.50
0 5 10 15 20
collimation
rela
tive
CTD
I plus4vol zoom
Single slice
Multi-slice
ImPACT EFOMP 2002 33
ImPACT 1996-1997 survey• Enabled NRPB datasets to be used for
99% of installed scanners in the UK
• New scanners are measured routinely by ImPACT and published
CTDI in air and in phantom ImF
ImPACT EFOMP 2002 34
• Dose Reference Levels CTDIw, DLP
EUR16262 Quality Criteria for CT
CTDIw =1/3CTDIc +2/3CTDIpc = centre positon, p= periphery position
ImPACT EFOMP 2002 35
• Name can be misleading– Siemens Plus, Plus 4 – Philips LX, AV Expander
• Filtration can change on existingscanner models – name stays the same– need to identify by model name & date
Scanner models
ImPACT EFOMP 2002 36
• Survey data tried to minimise errors– chambers cross checked– extensive work checking consistency of data
• data > 10% from mean of a group looked at
• newer scanners use our own data• checked against manufacturers data
• Data for 10 mm slices only• used for matching • look-up tables simplified• need focal spot with narrow slices
ImPACT data
ImPACT EFOMP 2002 37
ImPACT survey• Goal achieved
– matching new scanners to NRPB MC datasets• Additional use for phantom CTDI
– EUR16262 (10 mm data)• Data continuously updated
– Need to keep aware of manufacturers’ changesCTDIPxCTDIair
ImPACT EFOMP 2002 38
Acknowledgements
• N. Keat, M.A. Lewis, H. Kiremidjian, Dr. S. Sassi, Dr. A.J Britten,
• S E Jackson (DoH)• Department of Health Grant Ref RPP 26
• Dr. P. Shrimpton (nrpb)• Participating centres
ImPACT Imaging Performance
Assessment of CT Scanners
A Medical Devices Agency Evaluation Group
www.impactscan.org