Prediction of Abdominal Aortic Aneurysm Growth Using Dynamical Gaussian Process Implicit Surface.

Area of Science:

• Biomedical imaging
• Computational modeling
• Medical prediction

Background:

• Abdominal Aortic Aneurysm (AAA) growth prediction is crucial for patient management.
• Existing methods for AAA growth prediction have limitations in accuracy and patient-specificity.

Purpose of the Study:

• To introduce a novel computational approach for predicting future AAA growth using longitudinal CT scans.
• To quantify the uncertainty associated with AAA growth predictions.

Main Methods:

• Developed a Dynamical Gaussian Process Implicit Surface (DGPIS) model.
• Utilized Gaussian process regression for field construction from CT data.
• Learned a dynamic model to represent AAA surface evolution over time.

Main Results:

• Evaluated the method on a dataset of 7 subjects with multiple CT scans.
• Demonstrated superior prediction performance compared to simple extrapolation and Principal Component Analysis (PCA).
• Achieved accurate AAA surface prediction with quantified uncertainty.

Conclusions:

• Introduced a novel, patient-specific approach for predicting AAA growth and uncertainty.
• The method shows potential to guide clinical decisions in AAA treatment and monitoring.