Surgical Tracker Marker Geometry for Occlusion-Robust Pose Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional surgical navigation systems face challenges with high power consumption, occlusion issues due to moving objects and personnel, and confusion from multiple tracking features and light sources, leading to inaccurate tracking and increased complexity.
Innovation Solution
A surgical navigation system using a monocular camera and predefined marker geometries for robust marker assignment, combined with a peer-to-peer tracker network for redundancy and distributed processing, allowing efficient tracking even with reflections and light interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple cameras are used to track surgical objects, then tracking coverage and reliability are improved, but power consumption and system complexity increase
Solution Approach 1:
The system divides the tracking function across multiple independent tracker units, each with its own camera and processing capabilities. Each tracker segment handles a portion of the surgical workspace independently, allowing the system to maintain reliable tracking coverage while reducing the power burden on any single unit compared to a centralized multi-camera system.
Solution Approach 2:
The peer-to-peer tracker network dynamically assigns tracking responsibilities based on current workspace conditions. When a surgical object enters a tracker's field of view, that tracker automatically assumes responsibility for tracking it, and trackers communicate their status and capabilities in real-time. This dynamic allocation optimizes power consumption by activating only the necessary trackers rather than running all cameras at full power continuously.
2Measurement precision
If multiple cameras with overlapping fields of view are used, then tracking accuracy is improved, but the requirement for accurate relationship data between cameras increases system complexity
Solution Approach 1:
The system merges the coordinate systems of multiple trackers through peer-to-peer communication and automatic transformation. Each tracker maintains its own local coordinate system, but through shared communication of marker positions and relative transformations, the system seamlessly integrates data from multiple trackers into a unified surgical workspace coordinate system, eliminating the need for complex pre-calibration of inter-camera relationships.
Solution Approach 2:
Each tracker unit is self-sufficient with its own processing capabilities and can independently identify and track markers within its field of view. The trackers automatically perform self-calibration and coordinate transformation relative to neighboring trackers through peer-to-peer communication, eliminating the need for external calibration procedures and reducing system configuration complexity.
3Adaptability or versatility
If conventional navigation systems track multiple objects with fixed tracking features, then comprehensive tracking is achieved, but confusion from multiple features and light sources reduces tracking accuracy
Solution Approach 1:
The system uses markers with distinct visual characteristics including color variations to differentiate between multiple trackers and surgical objects. Each tracker and surgical instrument is equipped with uniquely identifiable markers that emit or reflect light in specific color patterns, allowing the camera system to automatically distinguish between multiple tracking features and filter out irrelevant light sources from the surgical environment.
Solution Approach 2:
The peer-to-peer tracker network implements continuous feedback communication where each tracker reports detected markers, their positions, and confidence levels to the navigation system. This feedback mechanism allows the system to resolve ambiguities when multiple trackers are present by cross-validating detections and eliminating false positives from reflected light or other干扰 sources, thereby maintaining high tracking accuracy across multiple objects.
4Ease of operation
If surgical personnel and objects move between camera and tracked object, then surgical workflow flexibility is improved, but occlusion causes tracking loss
Solution Approach 1:
The surgical workspace is divided into multiple overlapping fields of view covered by different tracker units. When one tracker's view is occluded by surgical personnel or objects, adjacent trackers continue to monitor the same surgical objects from different angles. This segmented coverage ensures that tracking continuity is maintained even when individual camera views are blocked, as long as at least one tracker maintains an unobstructed view.
Solution Approach 2:
The system dynamically switches tracking responsibility between different trackers based on real-time visibility conditions. When occlusion is detected by one tracker, the navigation system automatically transfers tracking of affected surgical objects to alternative trackers with clear views. This dynamic reassignment occurs transparently without interrupting the surgical workflow, maintaining both flexibility and tracking reliability.
Data Source
AI summary
Systems and methods for tracking an object in a workspace. A tracker is disposed relative to the object and includes a predefined geometry of markers having first and second marker arrangements, each including three markers in a collinear relationship. A camera with a view of the tracker generates an image comprising blobs, where a subset of the blobs is directly generated from the markers of the tracker, and a remainder is not. Blob groups each including three of the blobs and satisfying a collinear criterion are identified as candidates for having been directly generated by the marker arrangements. The subset of blobs is differentiated from the remainder of blobs based on the blob groups and the predefined marker geometry, and the differentiated subset of blobs is assigned to the markers of the tracker. The pose of the tracker in the workspace is then estimated based on the assigned blobs.