Factors affecting and computation of myocardial perfusion reference images
ABSTRACT Many quantitative analysis methods for myocardial perfusion studies require as a central step a comparison with a 'normal' or average density distribution map or reference image. It has been recognized, however, that the normal distribution can be affected by patient attributes, including sex and weight or body habitus, and by acquisition attributes, including the choice of tracer and the position of the patient during imaging. Some authors have proposed separate reference images for the sexes and the tracer. This approach fails if a large number of binary attributes have to be considered, since one would need 2" reference images for each attribute. The problem is compounded when continuous attributes (e.g. age and weight) are included, especially if the approach is to average separate homogeneous groups for each attribute. We propose to create case-specific reference images for the interpretation of myocardial perfusion studies by creating a model based on the influence of each attribute. From a non-homogeneous population of normal cases, or cases presumed to be normal on the basis of the Diamond and Forrester stratification, the effect of patient and study attributes on the density distribution in the stress image and the density differences between rest and stress images were computed. The effects are computed by multi-linear regression, to account for cross-correlation. Significance is assigned on the basis of a partial Fisher test. The data are myocardial perfusion images matched in 3D to a template by an elastic transformation. Even though there was some cross-correlation in the data, we were able to show independent effects of sex, position (prone or supine), age, weight, tracer combination and stress method (exercise, persantine and adenosine). Taken as a whole, the multi-linear regression demonstrated a significant effect in 72% of the pixels within the myocardial volume. In addition, the distribution predicted by the model was equivalent to average images from homogeneous matched groups. In conclusion, our approach makes it possible to produce case-specific reference images without the need for multiple homogeneous large groups to produce averages for each possible patient or study attribute.
- [Show abstract] [Hide abstract]
ABSTRACT: Viability and left ventricular ejection fraction (LVEF) are essential measures for the assessment of myocardial infarction (MI). These 2 variables may be evaluated simultaneously by means of thallium-201 gated single photon emission computed tomography (SPECT); however, the precision and accuracy of LVEF measurements with this isotope remain controversial, particularly in cases of extended perfusion defects and poor count densities. Fifty patients with a history of MI underwent a 20-minute rest and a 4-hour redistribution Tl-201 gated SPECT viability protocol, immediately followed by a technetium-99m planar equilibrium radionuclide angiography (ERNA). On gated SPECT images, various count statistics were calculated, and perfusion was automatically quantified by means of CardioMatch, which provided both the size and severity of MI defects. Rest and redistribution LVEFs were determined from gated SPECT with Germano's algorithm, whereas LVEFs were calculated from ERNA using the manufacturer's software. Mean LVEF values calculated with rest gated SPECT, redistribution gated SPECT, and planar ERNA were 30% +/- 13%, 30% +/- 13% and 33% +/- 13%, respectively. Significant differences between repeated gated SPECT LVEFs were not shown by means of the paired t test. Correlation coefficients were high between 20-minute and 4-hour scans (r = 0.89) and between gated SPECT and ERNA (r = 0.88 and r = 0.92 at 20 minutes and 4 hours, respectively). Additionally, close agreement between gated SPECT and ERNA was shown by means of the Bland-Altman plot, despite an underestimation of 3 units. Finally, neither the technical conditions (count density, heart rate, lung uptake, etc) nor the perfusion alteration (size, severity, redistribution) appeared to interfere with the precision and accuracy of gated SPECT LVEF measurement. Tl-201 gated SPECT is a precise method for assessing LVEF within the same patient at 4-hour intervals, even with a substantial count decay, and it gives accurate results compared with planar ERNA, even in the case of large perfusion defects.Journal of Nuclear Cardiology 12/2000; 8(1):31-9. DOI:10.1067/mnc.2001.109863 · 2.65 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Left ventricular ejection fraction (LVEF) and viability are essential variables for the prognosis of myocardial infarction and can be measured simultaneously by (201)Tl gated SPECT; however, most algorithms tend to underestimate LVEF. This study aimed to evaluate a new myocardial tracking algorithm, MyoTrack (MTK), for automatic LVEF calculation. A rest/redistribution (20 min/4 h) (201)Tl gated SPECT protocol followed immediately by a (99m)Tc equilibrium radionuclide angiography (ERNA) was performed in 75 patients with history of myocardial infarction. Quality of myocardial uptake was evaluated from count statistics and automatic quantification of defect sizes and severities (CardioMatch). LVEFs were calculated both with Germano's quantitative gated SPECT (QGS) algorithm and with MTK. Briefly, the originality of this algorithm resides in the unique end-diastole segmentation, matching to a template and motion field tracking throughout the cardiac cycle. ERNA LVEF averaged 33% +/- 14%. QGS significantly underestimated this value at 20 min (30% +/- 13%, P < 0.001) and at 4 h (30% +/- 13%, P < 0.0001). By contrast, MTK did not miscalculate LVEF at 20 min (34% +/- 14%, probability value was not significant) though a similar underestimation occurred at 4 h (31% +/- 13%, P < 0.02). Individual differences between early and late gated SPECT values and differences between gated SPECT and ERNA values did not correlate with the extension of perfusion defects, count statistics, or heart rate. MTK algorithm accurately calculates LVEF on early/high-count images compared with ERNA [corrected], even in patients with severe perfusion defects, but tends to underestimate LVEF on delayed/low-contrast images, as other algorithms do.Journal of Nuclear Medicine 06/2001; 42(6):845-52. · 5.56 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Assessment of perfusion defect extent is essential for determining prognosis after a myocardial infarction (MI), but quantification methods usually rely on segmental analysis, which may lack accuracy. We present an automated voxel-based and template-based approach for precise quantification of perfusion defect extent and reperfusion evolution. Coronary angiography and stress/reinjection (201)Tl tomography were performed prospectively on 49 patients with recent MI (45 men; mean age +/- SD, 54 +/- 10 y), before and 3 mo after revascularization (40 angioplasties and 9 bypasses). Perfusion defect extent was quantified using expert 16-segment visual scoring of the slices and a 3-dimensional (3D) method with spatial normalization between times 1 and 2. Briefly, the latter automatically extracted myocardial edges, matched them to a reference template, and compared the perfusion intensity in each voxel with the intensity of the corresponding voxel in a control population of 100 healthy subjects. Reocclusion occurred in 12 patients within 3 mo of surgery (all had undergone angioplasty). The perfusion gain between times 1 and 2, assessed by visual analysis, was significantly higher in permeable patients than in reoccluded patients: 12.4% +/- 13.3% and 2.3% +/- 8.2% of the initial stress defect, respectively (P = 0.02). Proportional gains, measured with the quantitative 3D method, were 4.5% +/- 3.6% and 1.9% +/- 2.7%, respectively (P = 0.02). Furthermore, the 3D method allowed measurement within the initial ischemic defect (reversible part of the stress defect at time 1), the extent of myocardium whose perfusion improved at time 2 (reperfusion), and the extent of myocardium whose perfusion remained unchanged (residual ischemia). A voxel-by-voxel analysis of these regions revealed that the proportion of reperfusion was significantly higher in permeable patients than in reoccluded patients: 60.0% +/- 21.3% versus 40.0% +/- 22.5%, respectively (P = 0.008). This was cumbersome to quantify using visual analysis and did not reach statistical significance, likely because of segmental division (partial-volume effect) and absence of spatial normalization. The 3D voxel-based quantification allows satisfying assessment of reperfusion 3 mo after MI. Moreover, the automated analysis using spatial normalization should facilitate a reproducible assessment of large populations over time.Journal of Nuclear Medicine 01/2005; 45(12):1981-8. · 5.56 Impact Factor