Evaluating the accuracy of planar gated blood pool processing software using simulated patient studies.

Area of Science:

• Nuclear Medicine
• Cardiovascular Imaging
• Medical Physics

Background:

• Planar gated blood pool (GBP-P) imaging is crucial for non-invasive assessment of left ventricular ejection fraction (LVEF).
• Accurate LVEF monitoring is vital for patients undergoing cardiotoxic chemotherapy.
• Concerns exist regarding the accuracy of commercial software for LVEF determination despite improved reproducibility.

Purpose of the Study:

• To create a database of GBP-P studies with known LVEF values using Monte Carlo simulations.
• To evaluate the accuracy of LVEF measurements from four commercial software programs.
• To assess the reproducibility of LVEF values derived from simulated GBP-P studies.

Main Methods:

• Utilized anthropomorphic 4D-XCAT models to generate 64 realistic GBP-P studies via Monte Carlo simulations.
• Processed simulated studies using four commercial software packages: Alfanuclear, Siemens, GE Xeleris, and Mediso Tera-Tomo.
• Assessed accuracy, intra-observer, and inter-observer reproducibility of LVEF measurements.

Main Results:

• Demonstrated a strong correlation between software-calculated LVEF and true LVEF values from simulations.
• Observed a slight underestimation of LVEF by all software programs at lower LVEF values.
• Reported excellent intra- and inter-observer reliability for LVEF measurements, indicating reduced subjectivity.

Conclusions:

• Commercial software programs accurately and reliably determine LVEF from simulated GBP-P studies.
• The observed underestimation bias at low LVEF necessitates careful interpretation to avoid excluding chemotherapy candidates.
• Future research should focus on mitigating underestimation biases and validating findings across diverse patient populations.