Kalman Filter 3D Tumor Tracking From 2D Projections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current radiation therapy methods for tumors in the thorax, abdomen, and pelvis lack real-time motion monitoring capabilities, especially during high-dose treatments, requiring additional and expensive equipment, and existing 2D-3D estimation algorithms are ill-posed and computationally expensive.
Innovation Solution
A Kalman filter framework is used to estimate 3D tumor position from 2D image projections on a standard linear accelerator, eliminating the need for an initial learning period and reducing computational costs, while being robust against measurement noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If additional dedicated equipment (e.g., CyberKnife, Calypso, RayPilot) is used for real-time motion monitoring, then measurement precision and reliability are improved, but device complexity and cost increase
Solution Approach 1:
The patent makes the standard linac kV imager perform dual functions: its primary function for imaging during treatment delivery and a secondary function for real-time tumor motion monitoring. By processing kV images acquired during routine treatment with appropriate algorithms (interdimensional correlation, probabilistic methods), the system enables 3D tumor position estimation without requiring dedicated motion monitoring equipment, thus achieving multi-functionality with existing hardware
Solution Approach 2:
The patent creates a virtual 3D model of tumor motion by processing 2D projection images from the standard kV imager. Through algorithmic reconstruction using interdimensional correlation and probabilistic estimation, the system generates accurate 3D position information that copies the functionality of dedicated tracking systems while using only the existing imaging hardware
2Device complexity
If 2D-3D estimation algorithms are used to estimate target position from kV images, then device complexity is reduced, but measurement precision deteriorates due to ill-posed nature of the problem
Solution Approach 1:
The patent introduces interdimensional correlation as an intermediary constraint that connects the 2D projection data to the 3D tumor position. By establishing mathematical relationships between motion in different dimensions (AP-LR correlation, AP-SI correlation), the algorithm transforms the ill-posed 2D-3D estimation problem into a well-constrained optimization problem, enabling accurate position estimation from limited 2D data
Solution Approach 2:
The patent implements iterative probabilistic estimation algorithms that use feedback from multiple 2D projections to progressively refine the 3D tumor position estimate. The system acquires kV images at multiple gantry angles, processes each projection through the estimation algorithm, and uses the accumulated information from all views to converge on an accurate 3D position, with the ability to update estimates in real-time as new projections become available
3Reliability
If real-time motion monitoring is implemented during high-dose treatments, then treatment reliability is improved, but loss of time increases due to additional imaging and processing
Solution Approach 1:
The patent implements continuous real-time motion monitoring throughout the entire treatment delivery process. The system acquires kV images continuously during treatment, processes them through the 2D-3D estimation algorithm in real-time, and provides continuous feedback on tumor position. This continuous monitoring ensures treatment reliability by detecting and accounting for any tumor motion that occurs during the treatment session, while the automated processing minimizes additional time requirements
4Device complexity
If standard linac kV imager is used instead of dedicated equipment, then device complexity and cost are reduced, but measurement precision deteriorates due to limited 2D information
Solution Approach 1:
The patent compensates for the limited 2D information from the standard kV imager by acquiring images at multiple gantry angles and using interdimensional correlation to infer the missing third dimension. The algorithm processes 2D projections from different viewing angles and uses the geometric relationships and motion correlations to reconstruct accurate 3D tumor position, effectively using angular diversity to overcome the inherent 2D limitation of the imaging system
Data Source
AI summary
An iterative Kalman Filter method was developed to address the need for estimating randomly moving targets during cancer radiotherapy on a standard equipped linear accelerator. Extensive evaluation of this method using different treatment scenarios shows sub-mm accuracy and precision. In addition, the system and method allows the target (or surrogates of the target) to be monitored without the need of a learning arc, reducing additional imaging dose to the patient. In addition, the method and system performs robustly against imaging and segmentation noise.