Video Respiration Rate Detection Using Neural Motion Fields
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for measuring respiration rate are prone to pathogen transmission and require physical presence, and remote measurement techniques suffer from poor accuracy in challenging conditions.
Innovation Solution
A camera-based system using convolutional neural networks (CNNs) processes video to calculate respiration rate by analyzing human torso movement, employing optical flow and segmentation masks to filter background noise, and converting average vector values to frequency domain for accurate rate determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If traditional in-person respiration rate measurement methods are used, then measurement accuracy is maintained, but pathogen transmission risk increases and patient travel time is required
Solution Approach 1:
The patent replaces direct physical contact measurement systems with an optical-based computer vision system. The respiration rate is measured by capturing video footage of the patient's torso and using image processing algorithms to detect chest wall movements, eliminating the need for physical contact between the patient and measurement device while maintaining measurement capability.
Solution Approach 2:
The patent introduces video footage as an intermediary medium between the patient and the measurement system. Instead of direct contact, the system captures optical information through a camera, processes the video data to extract respiration patterns, and derives respiration rate from these patterns, serving as a non-contact intermediary that prevents pathogen transmission.
2Object-affected harmful factors
If remote video-based respiration rate measurement is used, then pathogen transmission risk is reduced, but measurement accuracy deteriorates in challenging conditions
Solution Approach 1:
The patent performs preliminary actions by capturing multiple frames of video footage before processing. The system collects a sequence of images over time, ensuring sufficient data is gathered even in challenging conditions. This preliminary data collection allows the subsequent processing algorithms to work with adequate information for accurate respiration rate determination.
Solution Approach 2:
The patent maintains continuous useful action by processing video frames continuously throughout the recording period. The system analyzes multiple consecutive frames to track chest wall movements over time, ensuring that the measurement process continues uninterrupted and accumulates sufficient data points for reliable respiration rate calculation even when individual frames may be suboptimal.
3Measurement precision
If multiple processing steps are applied to improve accuracy, then respiration rate determination precision improves, but computational complexity increases
Solution Approach 1:
The patent segments the complex measurement task into distinct processing stages: video frame capture, chest wall movement detection through image differencing, vector field generation, and respiration rate calculation from movement patterns. This segmentation allows each stage to be optimized independently and simplifies the overall computational process while maintaining accuracy.
Solution Approach 2:
The patent implements self-service through the use of convolutional neural networks that automatically learn and adapt to different measurement conditions. The neural network is trained on diverse datasets including various lighting conditions, patient positions, and clothing types, enabling it to self-adjust and maintain high accuracy without requiring manual calibration or complex parameter tuning for each specific situation.
Data Source
AI summary
In some examples, a non-transitory computer-readable medium stores executable code, which, when executed by a processor, causes the processor to receive a video of at least part of a human torso, use a neural network to produce multiple vector fields based on the video, the multiple vector fields representing movement of the human torso, and determine a respiration rate of the human torso using the multiple vector fields.