Related Experiment Video
Updated: Jul 5, 2025
09:10
Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures
Published on: August 5, 2021
1.8K
BDIS-SLAM: a lightweight CPU-based dense stereo SLAM for surgery.
Jingwei Song1,2, Ray Zhang3, Qiuchen Zhu4
1United Imaging Research Institute of Intelligent Imaging, Beijing, 100144, China. jingweisong.eng@outlook.com.
International Journal of Computer Assisted Radiology and Surgery
|January 18, 2024
Summary
This study introduces BDIS-SLAM, a lightweight dense stereo simultaneous localization and mapping (SLAM) system for minimally invasive surgery (MIS). It achieves real-time performance on a single-core CPU, offering a low-cost solution for dense mapping in surgical procedures.
Area of Science:
- Computer Vision
- Robotics
- Medical Imaging
Background:
- Dense stereo simultaneous localization and mapping (SLAM) systems typically demand significant computational resources, limiting their application in resource-constrained environments like minimally invasive surgery (MIS).
- Existing SLAM approaches often require high-end parallel processing, which may not be available when computational power is allocated to other critical surgical tasks such as segmentation, detection, and tracking.
Purpose of the Study:
- To develop a lightweight dense stereo SLAM system capable of real-time performance on a single-core CPU for MIS.
- To overcome the limitations of high computational costs associated with conventional dense stereo SLAM methods in surgical applications.
Main Methods:
- Integration of a novel dense stereo mapping module, named BDIS-SLAM, with the ORB-SLAM2 system.
- Implementation of the Bayesian Dense Inverse Searching (BDIS) algorithm for real-time, CPU-level stereo matching.
- Development of a BDIS-based shape recovery and depth mosaic strategy integrated as a new thread for real-time stereo shape recovery.
Main Results:
- BDIS-SLAM demonstrated real-time performance exceeding 30 Hz on a modern single-core CPU in typical endoscopy/colonoscopy scenarios.
- The system showed minimal additional computational overhead compared to the base ORB-SLAM2 system.
- BDIS-SLAM successfully provided usable dense mapping for MIS on computationally constrained devices, despite simplifying processes by omitting deformation and fusion.
Conclusions:
- BDIS-SLAM offers a lightweight and efficient dense stereo SLAM solution for MIS, achieving real-time performance on single-core CPUs.
- This system presents a cost-effective approach for dense mapping in MIS, with potential applications in surgical robotics and augmented reality (AR) systems.
- The source code for BDIS-SLAM is publicly available, facilitating further research and development.
