Lesion-specific coronary artery calcium quantification for predicting cardiac event with multiple instance support...

Area of Science:

• Cardiovascular Imaging
• Artificial Intelligence in Medicine
• Biomedical Engineering

Background:

• Conventional whole-heart coronary artery calcification (CAC) quantification shows limitations in predicting near-term coronary events, particularly in high-risk individuals.
• Accurate prediction of near-term cardiac events remains a challenge in intermediate to high-risk populations.
• Existing methods may not fully capture the prognostic value of specific calcific lesions.

Purpose of the Study:

• To introduce a novel lesion-specific CAC quantification framework to enhance the near-term predictive value of CAC.
• To develop and validate a multiple instance support vector machines (MISVM) approach for improved cardiac event prediction.
• To characterize vulnerable or culprit lesions within coronary artery calcifications.

Main Methods:

• A lesion-specific CAC quantification framework was developed using geometric information, density, and clinical measurements.
• A multiple instance support vector machines (MISVM) model was constructed to predict cardiac events.
• The method was tested on data acquired using standard clinical imaging protocols on conventional CT scanners without hardware or protocol modifications.

Main Results:

• The proposed lesion-specific CAC quantification framework demonstrated significant improvement in predictive value compared to conventional methods.
• Receiver Operating Characteristic (ROC) analysis and net reclassification improvement (NRI) evaluation confirmed enhanced predictive accuracy.
• Leave-one-out validation supported the robustness and effectiveness of the MISVM model in identifying cardiac events.

Conclusions:

• Lesion-specific CAC quantification using MISVM offers a superior approach for predicting near-term cardiac events in intermediate to high-risk populations.
• The framework provides valuable insights into the characterization of vulnerable or culprit lesions in CAC.
• This novel method improves prognostic accuracy without requiring changes to existing CT hardware or imaging protocols.