British Journal of Radiology (2005) 78, 733-736
© 2005 British Institute of Radiology
doi: 10.1259/bjr/25662922
The effect of Butterworth and Metz reconstruction filters on volume and ejection fraction calculations with 99Tcm gated myocardial SPECT
T Vakhtangandze, MD, PhD
1,2
D O Hall, PhD
3
F V Zananiri, PhD
3 and
M R Rees, FRCP, FRCR
2
1 Department of Nuclear Medicine, Institute of Radiology and Interventional Diagnostics, Tbilisi, Georgia, 2 Department of Radiology, Bristol University, Bristol and 3 Department of Medical Physics and Bioengineering, United Bristol Healthcare Trust, Bristol, UK
Correspondence: Michael Rees, Department of Radiology, Bristol General Hospital, Guinea Street, Bristol BS1 6SY, UK
 |
Abstract
|
|---|
This study was carried out to measure the differences produced by change of reconstruction filter in calculations of left-ventricular end-diastolic volumes, end-systolic volumes, stroke-volumes and left-ventricular ejection-fractions from 99Tcm Sestamibi (Bristol-Myers Squibb) gated myocardial perfusion SPECT studies. 30 patients had gated SPECT myocardial perfusion imaging at rest. The acquired projections were separately filtered with two filters, a low-pass filter (Butterworth) and an edge-enhancement filter (Metz). Each study was then further processed to determine left-ventricular end-diastolic volume, end-systolic volume, stroke volume and ejection fraction, and to assess defect size. The results for each patient with the two filters were compared. Calculated end-diastolic volumes, end-systolic volumes and left-ventricular ejection fractions, for each filter, were well correlated. Stroke volumes showed worse correlation. The differences between left-ventricular ejection-fractions, end-diastolic volumes and end-systolic volumes were statistically significant. There was no significant difference in stroke volumes. Ejection fractions were inversely correlated with defect size, but change in ejection fraction due to filter was not. End-diastolic and end-systolic volumes were correlated with defect size, but change in volumes due to filter was not. Thus the results for changes produced by choice of filter are not dependent on defect size. Using different reconstruction pre-filters in gated myocardial perfusion SPECT significantly changes the results of calculations of physiological parameters. Each centre should be consistent in the use of filters as this may affect the clinical consequences of the result.
|
Introduction
|
|---|
Radionuclide myocardial perfusion imaging provides information on perfusion abnormalities, and is a sensitive test for diagnosing ischaemia. Electrocardiography (ECG) gated imaging provides additional information on global and regional myocardial contractile function [1], and allows the calculation of left ventricular end diastolic volume (EDV), end systolic volume (ESV), stroke volume (SV) and ejection fraction (LVEF). This functional information has been shown to give additional prognostic information [2, 3]. A number of methods have been proposed to calculate left-ventricular volumes, and hence to derive stroke volumes and ejection fractions [4–6]. In addition to the choice of calculation method, there are various acquisition and reconstruction parameters which may affect the quantitative results, such as whether the study was acquired with 8 or 16 frames per cycle [5, 7] and which filter was used in image reconstruction [8–10]. The accuracy of results may also be affected by patient-specific factors such as cardiac volume, patient size and perfusion defect size. It has been claimed that the effect of changing filters invalidates the clinical use of calculated ejection fractions [9].
This study was carried out to measure the differences produced by a change of reconstruction filter in calculations of left-ventricular EDV and ESV, SV and LVEF from 99Tcm-Sestamibi (Bristol-Myers Squibb, N. Billerica, MA) myocardial gated SPECT studies, using manufacturer-recommended acquisition parameters. This allowed us to assess the validity of measurements of left-ventricular volumes and ejection fractions in clinical practice.
|
Method
|
|---|
Patients
Resting 99Tcm-Sestamibi gated SPECT myocardial perfusion imaging was performed in 30 patients (16 male), mean age 65 years (range 43–82 years). All studies were carried out during 2002 and 2003 at the Bristol Royal Infirmary, UK.
Gated SPECT acquisition
Myocardial gated SPECT was carried out at rest, 30–45 min after intravenous injection of 400 MBq 99Tcm-Sestamibi. SPECT acquisition was carried out on a dual-head large field of view gamma camera (General Electric DST-XLi; General Electric, Amersham, UK), with the two heads placed at 90°. 32 projections (16 per head) were obtained in 64 x 64 matrices using a step and shoot acquisition over a 180° arc from right anterior oblique to left posterior oblique position. Acquisition zoom was 1.33, giving a pixel size of 6.77 mm. A single 20% energy window at 140 keV was used. 15 studies were acquired with 8 frames per cardiac cycle, and 15 with 16 frames per cardiac cycle, using an R-wave trigger and a 40% acceptance window, with 60 s of accepted counts per projection.
Gated SPECT data processing
Studies were processed on a General Electric Healthcare Vision PowerStation, with Vision release 5.2.1 software. Images were pre-filtered, and then reconstructed by filtered back-projection with a ramp filter. Two filters recommended by the manufacturer for reconstruction of gated SPECT studies, Butterworth order 4, cut-off frequency 0.25 cycles/pixel and Metz order 8, full-width half-maximum 4.0 mm, were compared. The former is a low pass filter [11], while the latter is an edge-enhancement filter [12]. Both these types of filter have been shown to be effective in reconstructing SPECT studies [13]. Myocardial end-diastolic volume (EDV, ml) and end-systolic volume (ESV, ml), stroke volume (SV, ml) and left-ventricular ejection-fraction (LVEF, %) were determined using a commercial semi-automatic gated SPECT processing software, Multidim (General Electric Healthcare), which uses the Stanford method [4]. All Multidim processing was performed by a single operator (TV) who processed each study using the two filters simultaneously, to minimize variability. The size of the perfusion defect was assessed using the programme Myoquant (General Electric Healthcare) [14].
Calculations
Statistical analysis was performed with the SPSS program version 11.5.0 for Windows (SPSS Inc., Chicago, IL). The Spearman rank correlation coefficient was used to test for correlations. The individual differences for each patient between filtering with Metz and Butterworth filters were calculated, and statistical differences were tested for using a paired t-test. The results for studies acquired with 8 and 16 frames per cardiac cycle were compared.
|
Results
|
|---|
The ESV and EDV, SV and LVEF are shown in Table 1
, calculated with Metz and Butterworth filters, and in Figure 1. The individual differences in results with change of filter are given in Table 1. There were no significant differences found between the studies acquired with 8 and 16 frames per cardiac cycle, so the results were combined. As can be seen from Figures 2 and 3, there is a good correlation between the results with the two filters.
View larger version (12K):
[in this window]
[in a new window]
|
Figure 3. Comparison of left-ventricular ejection-fraction (LVEF) calculated with Metz and Butterworth filters.
|
|
With the Metz filter rather than the Butterworth filter, left-ventricular EDV and ESV were significantly larger, the ejection fraction was significantly lower, and SV was not significantly changed (Table 1
). Spearman's rank correlation coefficients between results calculated with Metz and Butterworth filters were 0.96 for ESV, 0.88 for EDV, 0.94 for LVEF and 0.74 for SV.
Defect extent assessed by the Myoquant program ranged from 0 to 40%. As can be seen from Figure 4, the LVEF calculated with each filter is inversely related to Myoquant index, as expected [15], but the difference between results calculated with each filter showed no correlation with Myoquant Index. EDV and ESV correlated with Myoquant index, but differences between them did not (ESV results are shown in Figure 5). Thus it was reasonable to treat the patients as one group for the purposes of this study. The Multidim program worked for all patients, irrespective of defect size. It is recognized that accuracy must be compromised for hearts with large defects, due to lack of information on wall position in the defect area, but the difference in results calculated with different filters can still be determined.
|
Discussion
|
|---|
The Multidim program uses the Stanford method for calculating heart volume and ejection fraction from the gated SPECT images [4, 16]. In this method, the wall position is found from the centre of mass of the count rate distribution on a radius from the centre of the heart [4], which should be reasonably constant with reconstruction filter. The endocardial edge is found from the maximum slope of the profile between the cardiac centre and the wall centre [4, 17], which would be expected to vary with filter, and to give a thicker wall measurement with a smoother filter [18]. A thicker wall leads to a smaller cardiac volume, and so volumes were expected to be smaller with the smoother Butterworth filter, and larger with the sharper Metz filter. We would also expect that the difference in wall thickness would be similar at end-diastole and end-systole, so the SV would on average be little affected by change in filter, which we also found. In this case, the proportional difference in volume between filters would be greater at end systole than at end diastole. Hence the LVEF, which is calculated from 100 x (1–ESV/EDV)%, was lower with the Metz filter than with the Butterworth filter.
It is commonly assumed that cardiac volume calculation is strongly affected by defect size. We find here that the volumes and LVEF are correlated with defect size, as assessed by Myoquant index: this is probably a true correlation of defect size, cardiac volume and reduced LVEF, all related to poor cardiac function. However, the differences produced by change of filter were not strongly affected by defect size, so we can use these data to assess the effect produced by change of filter. In addition all measurements were made with the same processing technique which is semi-automatic, using the same data, so there was no variation in technique between the two filters.
Similar overall results were found in a previous study which used the same camera, computer system and software as used here to compare results obtained with a range of Butterworth filters [8], and in studies using another software package, quantitative gated SPECT (QGS), and a range of low-pass filters [5, 9]. A further study compared the Wiener edge-enhancement filter with the Butterworth filter in QGS [10] and found larger volumes and lower LVEF with the Wiener filter, similar to the relationship found with Metz and Butterworth in our study.
Ejection fraction has been established as a marker of prognosis particularly in heart failure [19] and after myocardial infarction [20]. It has also been quoted along with myocardial perfusion as a significant predictor of cardiovascular risk by the UK Drivers and Vehicles Licensing Authority [21]. Left ventricular volumes are measured by nuclear medicine routinely [22], and are now seen as a significant prognostic factor in cardiac disease. ESV has been found to be a major determinant of survival after myocardial infarction [23]. Normal ESV is 30–55 ml, an increase in EDV to 75 ml and 125 ml results in a 2.5, and 5 fold higher 5 year relative risk of cardiovascular death [24]. ESV is also the most sensitive parameter in determining improvement in left ventricular function after revascularization and may be the only indicator of improvement in these patients [25]. These measurements are also important in clinical trials and inclusion and exclusion criteria need to take account of differences in technique between centres in multi-centre trials. Given the importance of these parameters in prognosis, trials and in civil matters, i.e. whether an individual might retain a driving licence, it is vital that the determination of ejection fraction and left ventricular volume is accurate and the parameters by which they are calculated are clearly understood.
|
Conclusion
|
|---|
There were statistically significant mean differences in EDV and ESV and LVEF calculated using the two filters. There was no significant difference in SV. The differences found in our study were similar to those from previous studies comparing reconstruction with different filters [8–10].
The correlations between results obtained with the two filters were good for EDV and ESV and LVEF, while correlations for SV were lower, so SV calculated with gated SPECT should be used with care.
These results show that each centre should be consistent in their use of reconstruction filters in gated myocardial SPECT.
|
Footnotes
|
|---|
Dr Vakhtangandze was supported by a Fellowship Programme in Nuclear Medicine from the International Atomic Energy Agency. 
Received for publication October 19, 2004.
Revision received February 15, 2005.
Accepted for publication February 24, 2005.
|
References
|
|---|
- Anagnostopoulos C, Harbinson M, Kelion A, et al. Procedure guidelines for radionuclide myocardial perfusion imaging. Nucl Med Commun 2003;24:1105–19.[Medline]
- Beller GA, Zaret BL. Contribution of nuclear cardiology to diagnosis and prognosis of patients with coronary artery disease. Circulation 2000;101:1465–78.[Free Full Text]
- Jessup M, Brozena S. Heart failure. New Engl J Med 2003;348:2007–18.[Free Full Text]
- Goris ML, Thompson C, Malone LJ, Franken PR. Modelling the integration of myocardial regional perfusion and function. Nucl Med Commun 1994;15:9–20.[Medline]
- Germano G, Kiat H, Kavanagh PB, Moriel M, Mazzanti M, Su H-T, et al. Automatic quantification of ejection fraction from gated myocardial perfusion SPECT. J Nucl Med 1995;36:2138–47.[Abstract/Free Full Text]
- Faber TL, Cooke CD, Folks RD, Vansant JP, Nichols KJ, DePuey EG, et al. Left ventricular function and perfusion from gated SPECT perfuson images: an integrated method. J Nucl Med 1999;40:650–9.[Abstract/Free Full Text]
- Navare SM, Wackers FJT, Liu Y-H. Comparison of 16-frame and 8-frame gated SPET imaging for determination of left ventricular volumes and ejection fraction. Eur J Nucl Med 2003;30:1330–7.[CrossRef]
- Vera P, Manrique A, Pontvianne V, Hitzel A, Koning R, Cribier A. Thallium gated SPECT in patients with major myocardial infarction: effect of filtering and zooming in comparison with equilibrium radionuclide imaging and left ventriculography. J Nucl Med 1999;40:513–21.[Abstract/Free Full Text]
- Wright GA, McDade M, Martin W, Hutton I. Quantitative gated SPECT: the effect of reconstruction filter on calculated left ventricular ejection fractions and volumes. Phys Med Biol 2002;47:N99–N105.[Medline]
- Manrique A, Hitzel A, Gardin I, Dacher JN, Vera P. Impact of Wiener filter in determining the left ventricular volume and ejection fraction using thallium-201 gated SPECT. Nucl Med Commun 2003;24:907–14.[Medline]
- Germano G. Technical aspects of Myocardial SPECT imaging. J Nucl Med 2001;42:1499–507.[Abstract/Free Full Text]
- King MA, Schwinger RB, Doherty PW, Penney BC. Two-dimensional filtering of SPECT images using the Metz and Wiener filters. J Nucl Med 1984;25:1234–40.[Abstract/Free Full Text]
- Gilland DR, Tsui BMW, McCartney WH, Perry JR, Berg J. Determination of the optimum filter function for SPECT imaging. J Nucl Med 1988;29:643–50.[Abstract/Free Full Text]
- Benoit T, Vivegnis D, Foulon J, Rigo P. Quantitative evaluation of myocardial single-photon emission tomographic imaging: application to the measurement of perfusion defect size and severity. Eur J Nucl Med 1996;23:1603–12.[CrossRef][Medline]
- De Sutter J, Van de Wiele C, D'Asseler Y, et al. Automatic quantification of defect size using normal templates: a comparative clinical study of three commercially available algorithms. Eur J Nucl Med 2000;27:1827–34.[Medline]
- Goris ML, Boudier S, Briandet PA. Interrogation and display of single photon emission tomography data as inherently volume data. Am J Physiological Imaging 1986;1:168–80.
- Manoury C, Chen CC, Chua KB, Thompson CJ. Quantification of left ventricular function with Thallium-201 and Technetium-99m-Sestamibi myocardial gated SPECT. J Nucl Med 1997;38:958–61.[Abstract/Free Full Text]
- Hall DO, Zananiri FV. A phantom study of the effects of acquisition and reconstruction parameters on measured myocardial wall thickness and cavity size. Eur J Nucl Med 2002;Supplement 1, S355.
- Cintron G, Johnson G, Francis G, et al. Prognostic significance of serial changes in left ventricular ejection fraction in patients with congestive heart failure. Circulation 1993;87:17–23.
- Shah PK, Pichler M, Berman DS, Singh BN, Swan HJ. Left ventricular ejection fraction determined by radionuclide ventriculography in early stages of first transmural myocardial infarction. Relation to short-term prognosis. Am J Cardiol 1980;45:542–6.[CrossRef][Medline]
- http://www.dvla.gov.uk/drivers/medical/minutes_2004/meeting_minutes_28_apr_04.htm [Accessed 27 April 2005].
- Thorley PJ, Sheard KL, Rees MR. A comparison of methods for estimating left-ventricular volumes from radionuclide ventriculography. Physiological Measurement 1993;14:23–32.[Medline]
- White HD, Norris RM, Brown MA, Brandt PW, Whitlock RM, Wild CJ. Left ventricular end-systolic volume as the major determinant of survival after recovery from myocardial infarction. Circulation 1987;76:44–51.[Abstract/Free Full Text]
- Van Der Wall E, Bax JJ. Different imaging approaches in the assessment of left ventricular function: all things equal? Eur Heart J 2000;21:1295–7.[Free Full Text]
- Bax JJ, Arend FL, Schinkel AFL, Boersma E, Elhendy A, Rizzello V, et al. Effective left ventricular remodelling does not allow viable myocardium to improve in ejection fraction after revascularization and is associated with worse long term prognosis. Circulation 2004;110II-18–22.
Top
Abstract
Introduction
Butterworth filter 
. ESV, end systolic volume; EDV, end diastolic volume; SV, stroke volume.
) and end-systolic volume (
), compared with defect size measured by Myoquant Index.