SPECT versus planar gated blood pool imaging for left ventricular evaluation

Area of Science:

• Nuclear medicine imaging
• Cardiovascular imaging analysis
• Medical image processing

Background:

• A novel segmentation algorithm based on invariance of the Laplacian (IL) was developed.
• The algorithm aims to compute volumes and ejection fractions from medical imaging data.

Purpose of the Study:

• To evaluate the performance of the new IL algorithm.
• To compare its results against planar analysis and a standard algorithm (QBS) for accuracy in calculating left ventricular ejection fraction (LVEF).

Main Methods:

• Single photon emission computed tomography (SPECT) blood pool acquisition was performed in 202 patients.
• Left ventricular ejection fraction (LVEF) was calculated using the new IL algorithm, the standard QBS algorithm, and planar analysis (gold standard).
• Statistical analyses included correlation, Bland-Altman analysis, and linear regression to compare the methods.

Main Results:

• The IL algorithm showed a strong correlation with planar LVEF (r = 0.84), outperforming QBS (r = 0.77).
• Mean LVEF values were comparable across all methods (Planar: 32.72%, QBS: 32.32%, IL: 31.93%; P = .16).
• Bland-Altman analysis indicated negligible systematic bias for both 3D methods, with IL showing a smaller standard error of bias (7.44% vs. 9.36% for QBS) and no significant bias-LVEF correlation (P = .47).

Conclusions:

• The new IL segmentation algorithm provides results comparable to QBS and planar analysis for LVEF calculation.
• Unlike the QBS method, the IL algorithm demonstrated no correlation between the difference in LVEF and the magnitude of LVEF, suggesting improved accuracy across the spectrum of ejection fractions.