![LVEF](buttons/val_lvef_vbtn.gif)
![Perfusion](buttons/val_perfusion_vbtn.gif)
![Defect Size](buttons/val_defectsize_vbtn_p.gif)
![Reversibility](buttons/val_reversibility_vbtn.gif)
![Reproducibility](buttons/val_reproducibility_vbtn.gif)
![Prognostic Value](buttons/val_prognosticvalue_vbtn.gif)
|
|
Validation of 4D-MSPECT for the Estimation of Perfusion
Defect Size.
Objective: The purpose of this study was to evaluate
the quantitative perfusion algorithm in 3-D MSPECT for the estimation
of defect size in Tc-99m Sestamibi SPECT data acquired using two chest
phantoms, the Data Spectrum Elliptical Chest (DSEC) phantom with lung
and spine and the Capintec Heart (CAPH) phantom.
Methods: For both phantoms, the emission projection data
was acquired in a 15% window at 140 keV at 12 sec/step over 360° in 6°
steps. Low energy high resolution parallel hole collimators were used.
For the DSEC phantom, 11 defects (0, 5, 10, 20, 30 and 40ml) inserted
into the 120ml heart wall were imaged. Defects were located in the
anterior or posterior regions. For the CAPH phantom, 9 defects ranging
in size from 0 to 125 ml were inserted in the 178ml LV myocardial wall.
Tc-99m concentrations simulating a 111 MBq stress study were injected
in each of the phantoms. Polarmaps were constructed from the short axis
data using 4D-MSPECT. The polarmap data was compared to a normal male
database comprised of 36 low-likelihood patients. For each phantom,
defect size was recorded for defect thresholds ranging from 0.5 to 4.0
standard deviations (SD). Defect extent was compared to the known
defect volume using regression analysis and root mean square errors
(RMSE) were calculated.
Results: RMSE was minimized for a defect threshold of 2.5
to 3.0 SD for both phantoms. The fitted results for the DSEC phantom
were y=1.14x+2.16, RMSE=4.27 with correlation coefficient of r=0.96.
The results for the CAPH phantom were y=1.05x-9.90, RMSE=4.03 with a
correlation coefficient of r=0.99.
Conclusions: Based on the near unity correlation
coefficients, 4D-MSPECT demonstrated good linearity in estimating
defect size with only a slight overestimation of defect size (non-unity
slope values).
|