Home
Search
Collections
Journals
About
Contact us
My IOPscience
A new phantom for the assessment of nuclear medicine imaging equipment
This content has been downloaded from IOPscience. Please scroll down to see the full text. 1979 Phys. Med. Biol. 24 188 (http://iopscience.iop.org/0031-9155/24/1/019) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 128.6.218.72 This content was downloaded on 09/09/2015 at 11:42
Please note that terms and conditions apply.
PHYS. MED. BIOL., 1979, Vol. 24, No. 1, 188-192.
Printed in Great Britain
Technical Note
A New Phantom for the Assessment of Nuclear Medicine Imaging Equipment K. J. MURRAY, B . s c . , A. ~ T. ELLIOTT, PH.D.$ and J. WADSWORTH, M.SO.$
t Scientific and Technical Branch, D.H.S.S., 1 4 Russell Square, London WC1 , U.K. $Department of Nuclear Medicine, St. Bartholomew’s Hospital, London EC1, U.K. 8 Computing Unit for Medical Sciences, St. Bartholomew’s Hospital, London ECI, U.K. Received 13 February 1978, in$nal form 11 September 1978
1. Introduction The specification of Nuclear Medicine imaging equipment in terms of full width a t half the maximum of the line spread function (FWHM), modulation transfer function (MTF) and other definable physical quantities is to a certain extent an abstract exercise in that skilled interpretation of a large number of figures isrequired to appreciatetheir combined effect on the finalimage. Moreover, in many cases the data on every link in the chain from patient to hard copy are not available in standard compatible forms. These difficulties have led tothe use of variousphantomsto give avisualindication of performance. At present available phantoms suffer from a number of problems : the design described below is thoughtto overcome these.Themost commonly used phantomsarethe Anger and bar types.Theseare basicallyhigh contrast phantoms which represent an atypical clinical situation and emphasise certain aspects of camera performance at theexpense of others, e.g. intrinsic resolution as againstenergyresolution.To overcome this,phantoms such as that of Rollo (1976) have been made, but this hasa limited number of data points and is difficult to manufacture ; moreover, because spheres are used, the contrast is not constant. Results from these phantoms are not easily quantified and often represent an artificial situation. Similarproblemsarise in X-ray image intensifiers where the dose to the patient has to be balanced against image quality. While phantoms analogous to theAnger and bar phantoms are used, there arealso phantoms (Burger 1953, Hay 1960) which examine the resolution with respect to contrast in a similar fashion to that described here. Thephantom described below is designed to test the performance of a scintillation camera or scanner, including the display system, investigating the 0031-9155~79j010188+05 $01.00
@ 1979 The Institute of Physics
Assessment of Nuclear Medicine Imaging Equipment
189
resolution over a wide range of contrastlevels: sufficient data points are produced and presented in a way which allows easy manipulation and semiquantitative analysis of the results. It has been constructed as a cold spot phantom, since in clinical practice cold lesions are usually more difficult to detect than hot ones. The cold spots have been made square as this gives a further indication of image quality since squares are degraded intocircles at' thelimits of resolution but circles remain circles. 2.
Description
A diagram of the phantom is given in fig. 1. It consists of sets of square Perspex blocks of different sizes, each set of a given size having a range of heights. The whole array is enclosed in a Perspex box, the depth of which is equal t o the maximum height of the blocks so that when the box is filled with a 100
60
9
LO 10 20
% contrast
4
Fig. 1. Diagram of the phantom.
radioactive solution a range of contrast levels is available for each size. The base and walls of the box are 5 mm thick, with a 2 mm top. The square blocks have sides of 20, 17.5, 15, 12.5, 10, 8, 7 , 5 , 4 mm with 100, 80, 60, 40, 20, 10% contrast available in each size. The heights of the blocks were calculated to takeinto account self absorption in the radioactivesolutionfor 140 keV gamma photons. I n practice the phantom was split into four quadrants for ease of manufacture, handling andt o allow imaging on a standard field scintillation camera. I n use, each quadrant is filled from a common premixed solution t o ensure equal concentration in all four,and animage containing 400 k counts obtained. A typical camera image of the phantom is shown in fig. 2.
190
K. J . J i a r r a y et al.
If the image so obtained is examined a.nd the last squarevisible ineach row is noted, a set of coordina,tes (size, contrmt.) is obta,ined which ca.n be plotted. Moreover, if dat.a processing facilities are awilablc thedesign of the phantom with independent st.ep functions aIlows, with a cross-section plot, the absolute evaluation of contrast in t,hc ima.ge compared to the object, for various step widths. Although t o be strictly comparable to ~ I T Fmea.surements sinusoidal variations would of course berequired, manufacture would bevery complica.ted and in any ca.se practica.1 applica.t.ionof NTP is difficult.
1q.i~. 2.
L\
t,ypiCaI
CRIIICI’R
i~nt~pc.
3. Results
In order t.o check the consistency and reliability of the data, four observers were used, two of whom were trained and familiar with nuclear medicine images and two were completely inexperienced. Each observer was required to score the images (obtained from two machines, a Xuclear Enterprises Mk V H R camera and a Cleon scanner) as described above on four separate occasions separatedbydays and with no reference to their previous markings. The observers were requested to note the last square in each row which they werc quite confident they could see. The data were processed using the analysis of variance technique, the significance of the results being assessedby the F-test. It was shown that intra-observer differenceswerenegligible in all cases: over the four trials with the MK VHR camera, one of the trained observers was totally self consistent over the whole range of readings, one trained and one untrained observer had a single inconsistent reading and the other untrained observer had two inconsistent readings. There were slightly more inconsistencies with the Cleon scanner, particularly with one of the untrained observers. Inter-observer differences were present, particularly when the results fell in the 60-20% contrast range. There were,however, only minimal differences between the trained observers. Moreover, the inter-observer differences were
Assessment of Nuclear Medicine Imaging Equipment
191
consistent and in the same direction; thus one observer would mark consistently higher or lower than another but both were in agreement that a given image from one device was better or worse than that from another. Fig. 3 shows curves obtained by plotting the means of the data for the two trained observers (total of eight trials per machine). Inter-machine differences were shown to be significant and the major source of difference for size ranges 20-12.5 mm, where the Cleon was superior, and 8-4 mm, where the camera was
!L*
'351 -
'B ':
60
+
g
A
N E M k VHF,
x
Cleon
LO.
L%
____
:
19
20
20: I
l
E
@ 13 12 li 16 S m i l e s ! vlslble square l n m )
Fig. 3. Data means for two trained observers and the two machines.
superior. It should be noted that thecurve as shown is fitted to the datapoints but that the curve representingthe minimum detectablesize for a given contrast must lie just below that shown. Inter-machine differences are clearly demonstrated by this figure, and after carrying out an analysis of variance these differences were significant (except a t 4 mm). The two trained observers were in complete agreement on all trials a t 20, 17.5, 15, 12.5, 8 and 5 mm, and this emphasises the superiority of the Cleon over t'he range 20-12.5 mm and the camera over the range 8-5 mm. 4. Discussion
Personal preferences of a given observer, for example positive or negative images, low or high density or colour ranges, would of course affect the results. However, this phantom would allow a particular observer to examine in a critical manner variable parameters anddecide on those settingswhich give the best results in a sensitive and accurate fashion. Similarly, for a given machine, the phantom makes it possible to investigate the effect of total counts, background subtraction, photographic intensity, etc. t o ensure that t'he best possible images were being produced (Nicoll 1977). With a given setof machine parameters the phantom would allow accurate assessment of day to dayvariations for quality assurance. 5.
Conclusion
The phantom described offers the possibility of examining the performance of nuclear medicine imaging equipment over a clinically significant range. It
192
Assessment of Nuclear Medicine Imaging Equipment
allows comparison in a semi-quantitative manner between different machines as well as investigating theeffect of different parameters ina given machine and can be used for quality assurance, in all cases taking into account the whole chain between patient and film. The phantom is easy to make, either being milled from a thick sheet or fabricated. The agreement between observers is good though it is hoped to improve it further by the additionof contrast levels of 50 and 30% as itis in this range that discrepancies mostly occur. We therefore feel that it offers an easy, practical and realistic method of investigating the performance of imaging devices.
REFEREXCES BURGER, G. C. E., 1953, P h i l i p Technical Review, 11 ( l o ) , 79-96. HAY,G. A., 1960, Advances in Electronics and Electron Physics, 12, 363-377. NICOLL,J., J., 1977, B.Sc. Dissertation, University of Strathclyde. ROLLO, N. A., 1976, Quality Control for Scintillation Cameras, US Department of Health, Education and Welfare H E W Publication (FDA) 76-8046.
Search
Collections
Journals
About
Contact us
My IOPscience
A new phantom for the assessment of nuclear medicine imaging equipment
This content has been downloaded from IOPscience. Please scroll down to see the full text. 1979 Phys. Med. Biol. 24 188 (http://iopscience.iop.org/0031-9155/24/1/019) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 128.6.218.72 This content was downloaded on 09/09/2015 at 11:42
Please note that terms and conditions apply.
PHYS. MED. BIOL., 1979, Vol. 24, No. 1, 188-192.
Printed in Great Britain
Technical Note
A New Phantom for the Assessment of Nuclear Medicine Imaging Equipment K. J. MURRAY, B . s c . , A. ~ T. ELLIOTT, PH.D.$ and J. WADSWORTH, M.SO.$
t Scientific and Technical Branch, D.H.S.S., 1 4 Russell Square, London WC1 , U.K. $Department of Nuclear Medicine, St. Bartholomew’s Hospital, London EC1, U.K. 8 Computing Unit for Medical Sciences, St. Bartholomew’s Hospital, London ECI, U.K. Received 13 February 1978, in$nal form 11 September 1978
1. Introduction The specification of Nuclear Medicine imaging equipment in terms of full width a t half the maximum of the line spread function (FWHM), modulation transfer function (MTF) and other definable physical quantities is to a certain extent an abstract exercise in that skilled interpretation of a large number of figures isrequired to appreciatetheir combined effect on the finalimage. Moreover, in many cases the data on every link in the chain from patient to hard copy are not available in standard compatible forms. These difficulties have led tothe use of variousphantomsto give avisualindication of performance. At present available phantoms suffer from a number of problems : the design described below is thoughtto overcome these.Themost commonly used phantomsarethe Anger and bar types.Theseare basicallyhigh contrast phantoms which represent an atypical clinical situation and emphasise certain aspects of camera performance at theexpense of others, e.g. intrinsic resolution as againstenergyresolution.To overcome this,phantoms such as that of Rollo (1976) have been made, but this hasa limited number of data points and is difficult to manufacture ; moreover, because spheres are used, the contrast is not constant. Results from these phantoms are not easily quantified and often represent an artificial situation. Similarproblemsarise in X-ray image intensifiers where the dose to the patient has to be balanced against image quality. While phantoms analogous to theAnger and bar phantoms are used, there arealso phantoms (Burger 1953, Hay 1960) which examine the resolution with respect to contrast in a similar fashion to that described here. Thephantom described below is designed to test the performance of a scintillation camera or scanner, including the display system, investigating the 0031-9155~79j010188+05 $01.00
@ 1979 The Institute of Physics
Assessment of Nuclear Medicine Imaging Equipment
189
resolution over a wide range of contrastlevels: sufficient data points are produced and presented in a way which allows easy manipulation and semiquantitative analysis of the results. It has been constructed as a cold spot phantom, since in clinical practice cold lesions are usually more difficult to detect than hot ones. The cold spots have been made square as this gives a further indication of image quality since squares are degraded intocircles at' thelimits of resolution but circles remain circles. 2.
Description
A diagram of the phantom is given in fig. 1. It consists of sets of square Perspex blocks of different sizes, each set of a given size having a range of heights. The whole array is enclosed in a Perspex box, the depth of which is equal t o the maximum height of the blocks so that when the box is filled with a 100
60
9
LO 10 20
% contrast
4
Fig. 1. Diagram of the phantom.
radioactive solution a range of contrast levels is available for each size. The base and walls of the box are 5 mm thick, with a 2 mm top. The square blocks have sides of 20, 17.5, 15, 12.5, 10, 8, 7 , 5 , 4 mm with 100, 80, 60, 40, 20, 10% contrast available in each size. The heights of the blocks were calculated to takeinto account self absorption in the radioactivesolutionfor 140 keV gamma photons. I n practice the phantom was split into four quadrants for ease of manufacture, handling andt o allow imaging on a standard field scintillation camera. I n use, each quadrant is filled from a common premixed solution t o ensure equal concentration in all four,and animage containing 400 k counts obtained. A typical camera image of the phantom is shown in fig. 2.
190
K. J . J i a r r a y et al.
If the image so obtained is examined a.nd the last squarevisible ineach row is noted, a set of coordina,tes (size, contrmt.) is obta,ined which ca.n be plotted. Moreover, if dat.a processing facilities are awilablc thedesign of the phantom with independent st.ep functions aIlows, with a cross-section plot, the absolute evaluation of contrast in t,hc ima.ge compared to the object, for various step widths. Although t o be strictly comparable to ~ I T Fmea.surements sinusoidal variations would of course berequired, manufacture would bevery complica.ted and in any ca.se practica.1 applica.t.ionof NTP is difficult.
1q.i~. 2.
L\
t,ypiCaI
CRIIICI’R
i~nt~pc.
3. Results
In order t.o check the consistency and reliability of the data, four observers were used, two of whom were trained and familiar with nuclear medicine images and two were completely inexperienced. Each observer was required to score the images (obtained from two machines, a Xuclear Enterprises Mk V H R camera and a Cleon scanner) as described above on four separate occasions separatedbydays and with no reference to their previous markings. The observers were requested to note the last square in each row which they werc quite confident they could see. The data were processed using the analysis of variance technique, the significance of the results being assessedby the F-test. It was shown that intra-observer differenceswerenegligible in all cases: over the four trials with the MK VHR camera, one of the trained observers was totally self consistent over the whole range of readings, one trained and one untrained observer had a single inconsistent reading and the other untrained observer had two inconsistent readings. There were slightly more inconsistencies with the Cleon scanner, particularly with one of the untrained observers. Inter-observer differences were present, particularly when the results fell in the 60-20% contrast range. There were,however, only minimal differences between the trained observers. Moreover, the inter-observer differences were
Assessment of Nuclear Medicine Imaging Equipment
191
consistent and in the same direction; thus one observer would mark consistently higher or lower than another but both were in agreement that a given image from one device was better or worse than that from another. Fig. 3 shows curves obtained by plotting the means of the data for the two trained observers (total of eight trials per machine). Inter-machine differences were shown to be significant and the major source of difference for size ranges 20-12.5 mm, where the Cleon was superior, and 8-4 mm, where the camera was
!L*
'351 -
'B ':
60
+
g
A
N E M k VHF,
x
Cleon
LO.
L%
____
:
19
20
20: I
l
E
@ 13 12 li 16 S m i l e s ! vlslble square l n m )
Fig. 3. Data means for two trained observers and the two machines.
superior. It should be noted that thecurve as shown is fitted to the datapoints but that the curve representingthe minimum detectablesize for a given contrast must lie just below that shown. Inter-machine differences are clearly demonstrated by this figure, and after carrying out an analysis of variance these differences were significant (except a t 4 mm). The two trained observers were in complete agreement on all trials a t 20, 17.5, 15, 12.5, 8 and 5 mm, and this emphasises the superiority of the Cleon over t'he range 20-12.5 mm and the camera over the range 8-5 mm. 4. Discussion
Personal preferences of a given observer, for example positive or negative images, low or high density or colour ranges, would of course affect the results. However, this phantom would allow a particular observer to examine in a critical manner variable parameters anddecide on those settingswhich give the best results in a sensitive and accurate fashion. Similarly, for a given machine, the phantom makes it possible to investigate the effect of total counts, background subtraction, photographic intensity, etc. t o ensure that t'he best possible images were being produced (Nicoll 1977). With a given setof machine parameters the phantom would allow accurate assessment of day to dayvariations for quality assurance. 5.
Conclusion
The phantom described offers the possibility of examining the performance of nuclear medicine imaging equipment over a clinically significant range. It
192
Assessment of Nuclear Medicine Imaging Equipment
allows comparison in a semi-quantitative manner between different machines as well as investigating theeffect of different parameters ina given machine and can be used for quality assurance, in all cases taking into account the whole chain between patient and film. The phantom is easy to make, either being milled from a thick sheet or fabricated. The agreement between observers is good though it is hoped to improve it further by the additionof contrast levels of 50 and 30% as itis in this range that discrepancies mostly occur. We therefore feel that it offers an easy, practical and realistic method of investigating the performance of imaging devices.
REFEREXCES BURGER, G. C. E., 1953, P h i l i p Technical Review, 11 ( l o ) , 79-96. HAY,G. A., 1960, Advances in Electronics and Electron Physics, 12, 363-377. NICOLL,J., J., 1977, B.Sc. Dissertation, University of Strathclyde. ROLLO, N. A., 1976, Quality Control for Scintillation Cameras, US Department of Health, Education and Welfare H E W Publication (FDA) 76-8046.
