Machine Learning Patient Motion Monitoring in Radiotherapy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current medical imaging technologies for radiotherapy are inadequate in tracking patient anatomy accurately in real-time, especially during patient movement, leading to increased radiation exposure to healthy tissues and reduced effectiveness of treatment due to the inability to measure changing patient states directly.
Innovation Solution
A system utilizing machine learning and image processing to estimate patient states by generating a 3D image from partial measurements, such as 2D images, and using a patient model to predict future positions, allowing for more precise radiation therapy delivery by reducing unnecessary margins around the target.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If 2D imaging techniques are used to track patient anatomy, then the imaging process is simple and quick, but the tracking precision and completeness are insufficient
Solution Approach 1:
The patent transitions from 2D imaging to 3D imaging capabilities. The system acquires 3D medical images (such as CT or MRI scans) to capture the full volumetric anatomy of the patient, enabling accurate tracking of internal organs and tumors in three-dimensional space. This dimensional upgrade resolves the limitation of 2D imaging insufficiently tracking anatomical structures while maintaining clinical feasibility.
2Measurement precision
If surface information or markers are used to track patient state, then the technique is non-invasive, but the accuracy of internal patient state measurement is reduced
Solution Approach 1:
The patent uses image-guided registration techniques where 3D medical images serve as an intermediary between external tracking systems and internal anatomy. The system registers 2D projection images or surface marker data with 3D volumetric images to infer internal organ and tumor positions. This intermediary approach allows non-invasive external measurements to accurately reflect internal patient state by leveraging the geometric relationships captured in 3D imaging.
3Reliability
If large margins are added around radiation target to account for patient motion, then the effectiveness of radiation delivery is maintained, but the radiation exposure to healthy tissues increases
Solution Approach 1:
The patent implements dynamic tracking and adaptation of radiation delivery parameters during treatment. The system continuously monitors patient anatomy using 3D imaging and adjusts radiation beam positioning, intensity, and timing in real-time based on detected organ and tumor motion. This dynamic approach replaces static large margins with active compensation, maintaining treatment effectiveness while reducing unnecessary radiation exposure to surrounding healthy tissues.
4Productivity
If real-time patient state monitoring is implemented, then the precision of radiation delivery is improved, but the system complexity and processing requirements increase
Solution Approach 1:
The patent performs preliminary processing of 3D medical images before radiation treatment to create pre-planned radiation therapy schemes. The system pre-processes anatomical data, identifies target volumes and organs at risk, and establishes baseline registration frameworks. This preliminary action reduces the complexity of real-time monitoring during treatment, as the heavy computational lifting is completed beforehand, allowing faster real-time adjustments with reduced processing requirements.
Data Source
AI summary
Systems and techniques may be used to estimate a patient state during a radiotherapy treatment. For example, a method may include generating a dictionary of expanded potential patient measurements and corresponding potential patient states using a preliminary motion model. The method may include training, using a machine learning technique, a correspondence motion model relating an input patient measurement to an output patient state using the dictionary. The method may include estimating, using a processor, the patient state corresponding to an input image using the correspondence motion model.