Subject-specific models for image-guided cardiac surgery.

Area of Science:

• Medical Imaging
• Computational Anatomy
• Cardiovascular Surgery

Background:

• Minimally invasive cardiac surgery (MICS) lacks routine 3D visualization for planning and guidance.
• Subject-specific geometric models from 3D preoperative images can aid in planning port locations and procedure rehearsal.
• Intraoperative image registration enables guidance using these models during surgery.

Purpose of the Study:

• To develop a method for extracting subject-specific heart geometry from preoperative MR images for MICS.
• To overcome challenges posed by low-quality clinical imaging data (resolution, SNR, artifacts).
• To validate the accuracy and applicability of the generated dynamic models for surgical training and guidance.

Main Methods:

• A three-step approach: (1) generating a high-quality template model, (2) registering the template to preoperative MR data, and (3) animating the model throughout the cardiac cycle.
• Utilizing preoperative MR images despite their limitations in resolution, signal-to-noise ratio, and artifacts.
• Adapting the generated models using intraoperative ultrasound imaging for real-time surgical guidance.

Main Results:

• Dynamic, subject-specific heart models were successfully generated from low-resolution MR images (6 mm slices).
• The generated models achieved a mean error of 3.6+/-1.1 mm, demonstrating sufficient accuracy for MICS.
• The study confirmed the potential for adapting these models for intraoperative guidance in the operating room.

Conclusions:

• The developed method provides accurate dynamic, subject-specific heart geometry models from low-quality MR images.
• These models are suitable for training and procedural planning in minimally invasive cardiac surgery.
• The approach shows promise for intraoperative guidance, enhancing surgical precision and safety.