Derivation and external validation of a deep learning model to predict changes in coronary plaque burden

Area of Science:

• Cardiovascular Imaging and Diagnostics
• Artificial Intelligence in Medicine
• Medical Data Analysis

Background:

• Predicting coronary plaque burden changes is crucial but challenging in cardiovascular disease management.
• Accurate forecasting of plaque progression or regression aids in personalized treatment strategies.

Purpose of the Study:

• To develop and validate a deep learning model for predicting changes in percent atheroma volume (ΔPAV).
• To utilize intravascular ultrasound (IVUS) data for forecasting coronary plaque burden dynamics.

Main Methods:

• A deep learning model incorporating bidirectional Long Short-Term Memory (biLSTM) layers was developed.
• The model was trained and validated using data from the IBIS-4 and PACMAN-AMI datasets, analyzing intravascular ultrasound (IVUS) pullbacks.
• The model provides both classification of plaque progression/regression and estimation of ΔPAV.

Main Results:

• The model achieved high accuracy (0.85) and F1 scores (0.85) for predicting plaque progression and regression in the derivation dataset.
• External validation on the PACMAN-AMI dataset demonstrated strong predictive performance with an accuracy of 0.84 and an F1 score of 0.84.
• The model successfully identified plaque regression (mean ΔPAV -4.57±3.73) and progression (mean ΔPAV 4.02±3.55).

Conclusions:

• This study presents the first deep learning model for detecting changes in coronary plaque progression.
• The model analyzes the rate of plaque burden change between adjacent intravascular ultrasound frames.
• This approach offers a promising tool for monitoring cardiovascular disease and guiding therapeutic interventions.