Optical coherence tomography-based machine learning for predicting fractional flow reserve in intermediate coronary...

Area of Science:

• Cardiovascular Imaging
• Interventional Cardiology
• Artificial Intelligence in Medicine

Background:

• Fractional flow reserve (FFR) is crucial for assessing coronary artery stenosis severity.
• Intravascular optical coherence tomography (OCT) provides detailed anatomical information but not functional data.
• Predicting FFR from OCT imaging alone has not been previously explored.

Purpose of the Study:

• To investigate the feasibility of using machine learning with OCT data to predict FFR.
• To evaluate the diagnostic performance of an OCT-based machine learning-FFR model.

Main Methods:

• A machine learning model was developed using OCT and FFR data from 125 patients with left anterior descending artery lesions.
• Data was split into training (5:1 ratio) and testing sets.
• The OCT-based machine learning-FFR was compared to wire-based FFR for ischemia diagnosis (FFR ≤ 0.8).

Main Results:

• The OCT-based machine learning-FFR demonstrated strong correlation with wire-based FFR (r=0.853, P<0.001).
• Diagnostic performance in the testing group: 100% sensitivity, 92.9% specificity, 87.5% PPV, 100% NPV, and 95.2% accuracy.
• The model successfully predicted ischemia based on FFR thresholds.

Conclusions:

• Machine learning applied to OCT imaging can accurately predict FFR.
• This approach enables simultaneous acquisition of anatomical and functional data in a single procedure.
• OCT-based machine learning-FFR holds promise for optimizing treatment strategies for intermediate coronary artery stenosis.