Operator-less processing of myocardial perfusion SPECT studies

Area of Science:

• Medical Imaging
• Nuclear Cardiology
• Image Processing

Background:

• Myocardial perfusion single-photon-emission computed tomography (SPECT) is crucial for diagnosing coronary artery disease.
• Accurate image reconstruction and reorientation are vital for reliable interpretation of SPECT studies.
• Manual processing of SPECT data can be time-consuming and prone to variability.

Purpose of the Study:

• To develop and validate a fully automated algorithm for generating reoriented tomographic images from myocardial perfusion SPECT projections.
• To assess the success rate and reproducibility of the automated algorithm across different SPECT camera configurations and radiotracers.

Main Methods:

• The algorithm employs a three-module software approach: 2D feature extraction for reconstruction limits, filtered backprojection for transaxial image reconstruction, and image reorientation into short-axis views.
• Validation was performed on a large dataset of 350 rest 201Tl and 350 stress 99mTc-sestamibi studies acquired on single, dual, and triple-detector cameras.

Main Results:

• The complete automated processing sequence achieved a high success rate of 93.6% across all studies.
• Individual modules demonstrated strong performance, with myocardial boundary determination successful in 96.3% and reorientation in 97.2% of studies.
• No significant differences in success rates were observed between 201Tl and 99mTc-sestamibi imaging.

Conclusions:

• The developed automated algorithm for myocardial perfusion SPECT processing is highly effective and reproducible.
• This automated approach offers potential for significant time and cost savings in clinical and research environments.
• The algorithm enhances accuracy in myocardial perfusion SPECT analysis, aiding in better patient management and research outcomes.