A novel craniotomy simulation system for evaluation of stereo-pair reconstruction fidelity and tracking

Xiaochen Yang, Logan W. Clements, Rebekah H. Conley, Reid C. Thompson, Benoît M. Dawant, Michael I. MigaProceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE 2016 · DOI

TL;DR

This paper introduces a novel craniotomy simulation device that accurately mimics realistic brain deformations during surgery. This device is used to validate a stereo-pair camera system for tracking these deformations, showing promising accuracy for real-time intra-operative use in image-guided neurosurgery.

What Problem Does This Paper Address?

Image-guided neurosurgery (IGNS) relies on accurate pre-operative imaging, but the brain can significantly deform during surgery (brain shift), rendering these initial plans less reliable. Compensating for this shift requires tracking the brain's deformation in real-time using intra-operative data. While methods like laser range scanners (LRS) exist, they can disrupt the surgical workflow. Stereo-pair cameras, integrated into operating microscopes, offer a less intrusive alternative, but their accuracy for measuring cortical displacement needs rigorous validation. This paper addresses the need for a reliable method to evaluate the fidelity of stereo-pair based surface reconstruction and tracking systems for brain shift compensation.

Methods

The core of this work is a custom-built craniotomy simulation device. This device uses a flexible silicone membrane with drawn vascular patterns to simulate realistic brain deformations, including in-plane stretching and out-of-plane "sag." To validate their stereo-pair reconstruction system, the authors employed a ground-truth optical tracking system with a stylus.

The stereo-pair reconstruction pipeline involves:

  1. Camera Calibration: Using a standard checkerboard pattern and a user-friendly GUI for calibration.
  2. 3D Reconstruction: Employing epipolar geometry and pinhole models to compute intrinsic and extrinsic camera parameters. They utilize a Bouguet's algorithm for stereo rectification to align image planes and minimize distortions.
  3. Disparity Map Computation: Using block matching or semi-global block matching.
  4. 3D Point Cloud Generation: Reconstructing 3D points from the disparity map using OpenCV.

The validation process involved:

  • Acquiring baseline stereo images and creating a 3D point cloud.
  • Using the optically tracked stylus to record the real-world positions of fiducial points and visible vessel features.
  • Inducing deformations (stretch and sag) using the simulation device.
  • Repeating stereo image acquisition and stylus tracking for each deformed state.
  • Registering all datasets to a common space (baseline) and comparing the positions and displacements of landmarks measured by the stereo system against the ground-truth stylus measurements.

Datasets

The paper does not explicitly mention using pre-existing datasets for training or evaluation. Instead, it introduces a novel craniotomy simulation device as a benchmark for evaluating their stereo-pair reconstruction and tracking system. This device generates realistic deformations and allows for comparison against ground-truth measurements from an optical tracking system.

Key Insights and Ideas

  • The importance of realistic simulation: Developing a physical simulation device that accurately mimics brain shift behaviors (stretch and sag) is crucial for robustly validating intra-operative tracking systems.
  • Stereo-pair potential: Stereo-pair cameras, when properly calibrated and processed, show significant promise for capturing intra-operative cortical deformations with minimal disruption to the surgical workflow.
  • Trade-offs in intra-operative data acquisition: The paper highlights the inherent trade-off between the amount of information gathered, the impact on the surgical workflow, and the effectiveness of that information. Stereo-pair systems aim to strike a better balance than methods like LRS.
  • Validation is key: Rigorous comparison against independent, high-accuracy tracking systems is essential for establishing the reliability of new imaging and reconstruction techniques in medical applications.

Key Results

The study reports the following quantitative results:

  • Calibration Error: Approximately 0.3 pixels.
  • Mean Target Registration Error (Localization Error): Approximately 1 mm.
  • Mean Displacement Error: Slightly above 1 mm (specifically, 1.2 mm average difference in displacement measurement is mentioned in the abstract).
  • Optical Stylus Localization Error: Estimated to be approximately 0.5 mm.

These results suggest that the stereo-pair reconstruction technique is sufficiently accurate for estimating intra-operative displacements in near real-time without negatively impacting the surgical workflow.

Limitations and Future Work

Limitations: * The paper acknowledges that one target feature disappeared in the "stretch" state due to it moving out of view, which is a realistic scenario but limits direct comparison for that feature. * The authors mention that a more rigorous study is ongoing to further estimate the localization error at target points for their specific landmark and stylus configuration. * The current work focuses on evaluating the fidelity of stereo-pair reconstruction and tracking.

Future Work: * The authors plan to use their stereovision system as a substitute for LRS in their preoperative and intraoperative computational pipeline for brain shift correction. * Further investigation into the localization error of their system is planned.

Summary updated

Abstract

Brain shift compensation using computer modeling strategies is an important research area in the field of image-guided neurosurgery (IGNS). One important source of available sparse data during surgery to drive these frameworks is deformation tracking of the visible cortical surface. Possible methods to measure intra-operative cortical displacement include laser range scanners (LRS), which typically complicate the clinical workflow, and reconstruction of cortical surfaces from stereo pairs acquired with the operating microscopes. In this work, we propose and demonstrate a craniotomy simulation device that permits simulating realistic cortical displacements designed to measure and validate the proposed intra-operative cortical shift measurement systems. The device permits 3D deformations of a mock cortical surface which consists of a membrane made of a Dragon Skin® high performance silicone rubber on which vascular patterns are drawn. We then use this device to validate our stereo pair-based surface reconstruction system by comparing landmark positions and displacements measured with our systems to those positions and displacements as measured by a stylus tracked by a commercial optical system. Our results show a 1mm average difference in localization error and a 1.2mm average difference in displacement measurement. These results suggest that our stereo-pair technique is accurate enough for estimating intra-operative displacements in near real-time without affecting the surgical workflow.
BibTeX
@inproceedings{yang2016novel,
  author = {Xiaochen Yang and Logan W. Clements and Rebekah H. Conley and Reid C. Thompson and Benoît M. Dawant and Michael I. Miga},
  title = {{A novel craniotomy simulation system for evaluation of stereo-pair reconstruction fidelity and tracking}},
  booktitle = {Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE},
  year = {2016},
  doi = {10.1117/12.2217301},
}
Figure