Patient Motion Detection via Projection Image Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current motion detection systems struggle to reliably distinguish and separate body movements occurring at different body parts, especially in clinical settings, due to their complexity and computational cost, leading to noisy estimations and incorrect inferences.
Innovation Solution
A device and method using a simple binary motion/no-motion detection algorithm applied to projection images, generating multiple projection images from different viewpoints to separate motion detection signals corresponding to different body parts, allowing for accurate analysis without explicit motion estimation or body part segmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion estimation algorithms like optical flow are used to detect movement at pixel-level resolution, then measurement precision of movement location is improved, but device complexity and computational cost increase significantly
Solution Approach 1:
The patent divides the image into multiple projection images from different viewpoints, and further segments each projection image into multiple regions of interest (ROIs) corresponding to different body parts. This segmentation allows independent motion detection for each body part, achieving precise movement localization without requiring complex pixel-level motion estimation algorithms across the entire image.
Solution Approach 2:
The patent applies different motion detection strategies to different regions (body parts) based on their specific characteristics. By identifying which body parts are moving and focusing computation only on those regions, the system achieves high measurement precision for movement detection while reducing overall computational complexity compared to processing the entire image uniformly.
2Measurement precision
If motion estimation algorithms with fine granularity are used, then measurement precision of movement is improved, but reliability decreases due to noise in clinical video recordings
Solution Approach 1:
By segmenting the image into projection images from different viewpoints and further into body part regions, the patent reduces the impact of noise. Each segmented region can be processed independently with simpler algorithms, and the segmentation itself acts as a form of noise filtering by focusing on specific anatomical structures rather than pixel-level variations.
Solution Approach 2:
The patent transforms the 2D image into multiple projection images representing different 3D viewpoints. This dimensional transformation provides geometric context that helps distinguish true movement from noise, as noise does not consistently appear across different projection angles while actual body movements do.
3Device complexity
If simple frame difference or correlation methods are used for motion detection, then device complexity is reduced, but measurement precision and ability to separate co-occurring movements deteriorates
Solution Approach 1:
The patent segments the image into multiple projection images from different viewpoints and further divides each into regions corresponding to specific body parts. This segmentation enables simple frame difference or correlation methods to be applied to each region independently, allowing the system to achieve precise movement location and separation of co-occurring movements without using complex algorithms.
Solution Approach 2:
By creating multiple projection images from different viewpoints, the patent adds a dimensional aspect that provides spatial context. This allows simple motion detection algorithms to achieve better precision by analyzing movement patterns across multiple projected views rather than in a single 2D image.
4Productivity
If motion detection is performed on the entire image, then productivity is maintained, but loss of information occurs regarding which specific body parts are moving
Solution Approach 1:
The patent segments the image into projection images and further into body part regions, allowing motion detection to be performed on each segment independently. This segmentation preserves body part information while maintaining processing efficiency, as each segment can be processed in parallel and the results are easily combined to provide comprehensive movement analysis.
Data Source
Figure 1
Figure 2
Figure 3A~3F
AI summary
The present invention relates to a device for detecting motion of a person. The present invention further relates to a system incorporating the aforementioned device and a corresponding method. The proposed device comprises a projection sweep unit configured to obtain image data comprising depth information corresponding to an image showing the patient and to generate one or more projection images of the patient for different viewpoints, a motion detection unit configured to detect motion of the patient in the one or more projection images and to generate one or more motion detection signals indicating if motion is detected or not in a respective projection image, and an analysis unit configured to detect and localize body movements of one or more body parts of the patient based on one or more combinations of motion detection signals.