Author: Debarghya China, Junghoon Lee, Ali Uneri 👨🔬
Affiliation: Department of Radiation Oncology and Molecular Radiation Sciences, Johns Hopkins University, Department of Biomedical Engineering, Johns Hopkins University, Johns Hopkins Univ 🌍
Purpose: This study aims to develop a population-based cardio-respiratory motion model and apply it to patient-specific 3D CTA to simulate 4D CTA and 2D+t fluoroscopy sequences. The developed motion model facilitates accurate guidance through real-time tracking of dynamic vessel motion during percutaneous coronary interventions (PCI).
Methods: The proposed framework integrates a population-based cardiac and respiratory motion model. The cardiac motion model was derived from a 4D public dataset using Principal Component Analysis to capture key cardiac dynamics. The public dataset was aligned spatially, temporally, and across cardiac phases, enabling the capture of cardiac motion throughout the End Diastolic (ED)-to-ED cycle. The respiratory motion model was generated using patient-specific respiratory parameters such as respiratory cycle, inhalation-to-exhalation ratio, number of phases, maximum motion amplitude, and directions of motions. Finally, the two models combined to create a comprehensive cardio-respiratory motion model, leveraging ECG and respiratory surrogates.
Results: Evaluations were performed using Dice similarity scores and volumetric metrics to assess the performance of the simulated motion model. The Dice scores and the volumes were calculated to measure the structural and volumetric alignment between the segmented left ventricle in the simulated 4D CTA and the ground truth from the public 4D dataset. The results showed that the left ventricle volume changes in the simulated model closely matched the patient-specific data, with all measurements consistently falling within the 95% confidence interval, demonstrating the model’s accuracy in capturing cardiac dynamics.
Conclusion: The results demonstrate the effectiveness of the proposed model in accurately simulating cardio-respiratory motion and simulating realistic 4D CTA and 2D+t fluoroscopy. The primary use of this population-based model is real-time tracking of 3D coronary artery and catheter/guidewire tip motion during PCI. Ongoing work aims to tune the proposed populated-based model with patient-specific motion through on-the-fly 3D/2D+t registration.