Standardized Radiomics Analysis of Clinical Myocardial Perfusion Stress SPECT Images to Identify Coronary Artery...

Area of Science:

• Cardiovascular imaging
• Radiomics
• Machine learning

Background:

• Myocardial perfusion (MP) stress single-photon emission computed tomography (SPECT) is a standard test for coronary artery disease (CAD).
• Coronary artery calcification (CAC) scoring via CT provides incremental diagnostic information but is less commonly reimbursed or deployed.
• There is a need for complementary information from widely used MP SPECT scans.

Purpose of the Study:

• To investigate the potential of radiomics analysis from normal MP stress SPECT scans to predict CAC scores.
• To assess if radiomics features can provide complementary diagnostic information to standard MP SPECT.

Main Methods:

• 428 patients with normal MP stress SPECT scans were analyzed.
• 487 radiomic features were extracted using the SERA software package.
• Multivariate machine learning models were built using radiomic features, clinical features, or a combination thereof to predict CAC scores.

Main Results:

• Unsupervised feature selection reduced 487 radiomic features to 56.
• No single radiomic feature directly correlated with CAC scores in univariate analysis.
• Combining radiomic features with clinical data significantly enhanced the prediction model for CAC scores across all cardiac segments.

Conclusions:

• Standardized radiomics analysis of MP stress SPECT, when combined with clinical features, can significantly predict CAC scores.
• This approach suggests that radiomics analysis can add diagnostic or prognostic value to standard MP SPECT.
• The findings support the potential for wider clinical adoption of radiomics with MP SPECT.