Radiation Fluoroscopy Marker Tracking via Learning-Based Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current radiation therapy systems face challenges in accurately tracking markers within the body due to the movement of patients and tumors, particularly with non-spherical markers that deform or rotate, leading to increased preparation times and reduced irradiation accuracy.
Innovation Solution
A radiation fluoroscopy apparatus that includes an image generation element, local structure detection, device determination, device location acquisition, and device tracking elements, allowing for the detection and tracking of markers without the need for pre-generated templates, regardless of marker shape or orientation, and minimizes the impact of static structures like bones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If template matching is used to detect markers, then detection accuracy is improved, but preparation time increases and the system cannot handle deforming markers
Solution Approach 1:
The patent replaces the mechanical template matching process with a learning-based detection system. The system learns marker characteristics from training images and automatically detects markers in fluoroscopic images without requiring manual template preparation, thereby eliminating time-consuming template matching while maintaining detection accuracy.
Solution Approach 2:
The system performs preliminary learning during a training phase where it acquires marker characteristics from training images taken before treatment. This preliminary action allows the system to be pre-configured with marker-specific detection parameters, enabling rapid marker detection during actual treatment without time-consuming real-time template matching.
2Measurement precision
If multiple templates are prepared for different marker orientations, then detection accuracy is improved, but device complexity and preparation time increase
Solution Approach 1:
The patent implements a universal learning-based detection system that can handle markers of various shapes, sizes, and orientations with a single detection algorithm. The system learns diverse marker characteristics during training and applies this knowledge to detect any marker variation during treatment, eliminating the need for multiple specialized templates for different orientations.
Solution Approach 2:
The system dynamically adjusts detection parameters based on learned marker characteristics rather than using fixed templates. The learning process extracts variable parameters such as marker size, shape, and orientation from training images, allowing the detection algorithm to adapt to different marker configurations without requiring separate templates for each parameter combination.
3Speed
If traditional fluoroscopy is used to track markers, then real-time tracking is achieved, but accuracy decreases when markers deform or rotate
Solution Approach 1:
The patent implements a dynamic detection system that continuously adapts to marker changes during treatment. The learning-based algorithm can handle real-time variations in marker shape, size, and orientation by comparing current fluoroscopic images against learned characteristics, maintaining both real-time tracking speed and accuracy even when markers deform or rotate due to patient movement or breathing.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach reduces patient preparation time, enhances irradiation accuracy, and improves the throughput of radiation therapy by enabling accurate tracking of markers with varying shapes and orientations without the need for multiple templates, thus improving the efficiency of radiation therapy.
Implementation Method 1
an image generation element that generates an image including an embedded marker inside the body of the subject based on a transmitted X-ray
Data Source
AI summary
A radiation fluoroscopy apparatus detects a marker and includes a control element, an image generation element 61 that generates an image including an embedded marker inside the body of the subject based on a transmitted X-ray. A device candidate detection element 62 detects the candidate of the marker, the local structure detection element 63 detects the local structure in the target region in a proximity of the candidate point of the marker, the device determination element 64 determines whether the local structure is the device such as the marker or not, the device location acquisition element 66 acquires the gravity center coordinate of the local structure, and the device tracking element 67 tracks the marker based on the location of the marker in each frame.