Augmented reality visualization of biomechanical wall stresses on abdominal aortic aneurysms using artificial...

Area of Science:

• Medical imaging and visualization
• Biomechanical engineering
• Artificial intelligence in healthcare

Background:

• Increasing medical imaging use for an aging population necessitates better patient comprehension.
• Clinicians face time constraints in processing and presenting 2D/3D diagnostic images to patients.
• Current methods lack efficient tools for visualizing complex biomechanical data like aneurysm wall stress.

Purpose of the Study:

• To demonstrate augmented reality (AR) for visualizing abdominal aortic aneurysms (AAAs).
• To integrate an artificial intelligence (AI) engine for predicting AAA wall stress.
• To develop a clinical workflow for accessible biomechanical status visualization.

Main Methods:

• Utilized an AI engine trained on 274 patient cases for stress prediction.
• Implemented automated segmentation of medical images to create 3D aneurysm models.
• Developed a pipeline for Microsoft HoloLens 2 to display 3D models with stress heatmaps.
• Processed data on a local server for point cloud generation and stress prediction.

Main Results:

• Neural networks and ensemble boosted trees accurately predicted wall stresses compared to finite element analysis.
• The AR system enabled visualization of biomechanical status without requiring imaging experts.
• The pipeline successfully generated 3D models with colormaps for AR viewing.

Conclusions:

• AR visualization of AAA biomechanics is feasible and enhances patient-physician communication.
• The AI-driven workflow offers an accessible method for understanding complex medical data.
• The technology is adaptable to various AR/virtual reality hardware systems.