Related Experiment Videos
Automatic quantification of left ventricular ejection fraction from gated blood pool SPECT
S D Van Kriekinge1, D S Berman, G Germano
1Department of Medicine, Burns and Allen Research Institute, Cedars-Sinai Medical Center, Los Angeles, Calif 90048, USA.
Summary
An automated algorithm accurately measures left ventricular ejection fraction (LVEF) using gated blood pool SPECT (GBPS). This method offers a reliable alternative to conventional techniques for cardiac imaging analysis.
Area of Science:
- Nuclear Cardiology
- Medical Imaging Analysis
- Computational Anatomy
Background:
- Gated blood pool single photon emission computed tomography (GBPS) offers superior cardiac chamber separation over planar radionuclide ventriculography (PRNV).
- Quantitative assessment of left ventricular ejection fraction (LVEF) is crucial for cardiac function evaluation.
Purpose of the Study:
- To develop and validate a fully automated algorithm for quantitative LVEF measurement from 3D GBPS image volumes.
- To establish a basis for advanced 3D analysis of cardiac wall motion.
Main Methods:
- An ellipsoidal coordinate system was defined for the left ventricle.
- Endocardial surface dynamics were computed using count data, gradients, and temporal Fourier transform of 8-interval GBPS data.
- LVEF was calculated from end-diastolic and end-systolic volumes derived from the dynamic surfaces.
Main Results:
- The automated algorithm achieved successful left ventricular identification and contouring in 70% of all studies and 85% in the validation group.
- A high correlation (r=0.89) was observed between LVEF measured by the automated GBPS algorithm and conventional PRNV.
- GBPS-derived LVEF values were significantly higher than PRNV values (average difference 2.8%, P < .004).
Conclusions:
- The developed automatic algorithm provides quantitative LVEF measurements that agree well with conventional radionuclide methods.
- This automated approach facilitates 3D analysis of cardiac wall motion, enhancing diagnostic capabilities.