Integrated system for point cloud reconstruction and simulated brain shift validation using tracked surgical microscope

Xiaochen Yang, Logan W. Clements, Ma Luo, Saramati Narasimhan, Reid C. Thompson, Benoît M. Dawant, Michael I. MigaProceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE 2017 · DOI

TL;DR

This paper presents an integrated system that uses a tracked surgical microscope to reconstruct 3D point clouds of the brain surface. This system aims to address the challenge of "brain shift" during neurosurgery by providing a cost-effective way to capture intraoperative deformations, offering a practical alternative to existing, more disruptive methods.

What Problem Does This Paper Address?

Image-guided surgery (IGS) systems are crucial for neurosurgery, but they struggle with "brain shift" – the deformation of brain tissue that occurs during surgery due to factors like fluid drainage or tissue manipulation. This shift can render pre-operative imaging data inaccurate, compromising surgical guidance. While solutions like intraoperative MRI, CT, or ultrasound exist, they have drawbacks such as radiation exposure, high cost, or poor image quality. The authors propose using sparse data from a stereoscopic microscope, a tool already present in many operating rooms, to compensate for brain shift. The challenge is to effectively integrate this microscope into a workflow that allows for accurate 3D reconstruction and tracking of its movements to capture these deformations.

Methods

The core of this work is an interactive system built around a Zeiss OPMI Pentero surgical microscope equipped with two CCD cameras. The system enables:

  • Stereo Image Acquisition: Capturing stereo image pairs from the microscope.
  • 3D Point Cloud Reconstruction: A pipeline that includes camera calibration, image rectification, disparity computation (using Block Matching or Semi-Global Block Matching algorithms), and finally, 3D point cloud generation. A user-friendly GUI allows for parameter tuning and visualization.
  • Microscope Tracking: An optical tracking system (Polaris Spectra) is used with a rigid body attached to the microscope. This allows for real-time tracking of the microscope's position and orientation in 3D space.
  • Registration and Displacement Calculation: Point clouds acquired at different microscope positions are registered into a common coordinate system. This registration, combined with the tracked microscope data, allows for the computation of cortical surface displacements, effectively measuring brain shift.
  • Simulated Brain Shift: A mock craniotomy device is used to generate realistic deformations (stretch and sag) to simulate brain shift for validation.

Datasets

The paper does not explicitly mention training datasets. For evaluation, they used:

  • Calibration Phantom: A phantom with known geometry and fiducial disks was used for calibrating the stereo cameras and validating the microscope tracking.
  • Simulated Craniotomy Device: This device, with marked vessel features on a membrane, was used to simulate brain shift deformations. The ground truth for these deformations was captured using an optically tracked stylus.

Key Insights and Ideas

  • Leveraging Existing Tools: The system demonstrates that a standard surgical microscope, when augmented with tracking capabilities, can be a powerful tool for capturing intraoperative deformations, reducing the need for specialized, expensive equipment.
  • Integrated Workflow: The development of an interactive GUI that integrates image acquisition, reconstruction, and parameter tuning is crucial for making such a system practical for clinical use, even for users without extensive computer vision expertise.
  • Tracking is Key: The ability to accurately track the microscope's position is fundamental to registering point clouds acquired from different viewpoints and thus measuring displacements.
  • Simulated Deformations for Validation: Using a mock craniotomy device to create controlled, realistic brain shift scenarios is an effective way to validate the system's performance before clinical deployment.

Key Results

  • Reconstruction Accuracy: The stereo-pair camera calibration yielded an average epiline error of 0.2 pixels and a stereo projection error of 0.41 pixels. The reconstruction error of the calibration phantom was estimated to be approximately 1 mm.
  • Registration Accuracy: When registering point clouds from different microscope positions, mean distance errors of 0.64 mm (x-axis), 0.89 mm (y-axis), and 2.92 mm (z-axis) were reported, attributed to reconstruction and tracking errors.
  • Brain Shift Measurement Accuracy: The system achieved an average displacement error of approximately 2 mm when compared to the ground truth provided by the optical tracking system for simulated brain shift deformations.

Limitations and Future Work

  • Phantom Limitations: The current calibration phantom has sharp edges, which can lead to reconstruction errors. The authors suggest using a phantom without sharp edges (e.g., a sphere phantom) for improved results.
  • Fixed Microscope Settings: The current calibration method is limited to fixed microscope settings (focal length and zoom factor). Future work should aim to develop methods that can handle multiple microscope settings throughout a procedure.
  • Reconstruction Fidelity: Refinements in stereo reconstruction techniques, particularly for disparity computation and rectification, could lead to more accurate calibrations.
  • Error Sources: The authors acknowledge that errors can stem from stereo reconstruction, manual fiducial digitization, calibration phantom design, tracking system errors, and rigid body attachment issues.

The paper concludes that their integrated system is functional, user-friendly, and capable of gathering valid cortical data for measuring and compensating for brain shift, thereby extending the capabilities of conventional IGS systems.

Summary updated

Abstract

Intra-operative soft tissue deformation, referred to as brain shift, compromises the application of current imageguided surgery (IGS) navigation systems in neurosurgery. A computational model driven by sparse data has been used as a cost effective method to compensate for cortical surface and volumetric displacements. Stereoscopic microscopes and laser range scanners (LRS) are the two most investigated sparse intra-operative imaging modalities for driving these systems. However, integrating these devices in the clinical workflow to facilitate development and evaluation requires developing systems that easily permit data acquisition and processing. In this work we present a mock environment developed to acquire stereo images from a tracked operating microscope and to reconstruct 3D point clouds from these images. A reconstruction error of 1 mm is estimated by using a phantom with a known geometry and independently measured deformation extent. The microscope is tracked via an attached tracking rigid body that facilitates the recording of the position of the microscope via a commercial optical tracking system as it moves during the procedure. Point clouds, reconstructed under different microscope positions, are registered into the same space in order to compute the feature displacements. Using our mock craniotomy device, realistic cortical deformations are generated. Our experimental results report approximately 2mm average displacement error compared with the optical tracking system. These results demonstrate the practicality of using tracked stereoscopic microscope as an alternative to LRS to collect sufficient intraoperative information for brain shift correction.
BibTeX
@inproceedings{yang2017integrated,
  author = {Xiaochen Yang and Logan W. Clements and Ma Luo and Saramati Narasimhan and Reid C. Thompson and Benoît M. Dawant and Michael I. Miga},
  title = {{Integrated system for point cloud reconstruction and simulated brain shift validation using tracked surgical microscope}},
  booktitle = {Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE},
  year = {2017},
  doi = {10.1117/12.2255777},
}
Figure