Radar Heart Rhythm Detection Using Deep Learning Spectrograms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current radar-based vital sign detection methods face uncertainties due to unknown body conditions, sensor shaking, and environmental interferences, which affect reliable heart rhythm detection in real-world settings.
Innovation Solution
A deep learning algorithm using Short-Time Fourier Transform (STFT) to convert raw signals into two-dimensional image information, combined with a neural network that filters noise and identifies the relationship between heartbeat and non-heartbeat frequencies, employing high-pass and band-pass filtering to improve signal quality and enhance heart rhythm detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional algorithmic processing methods are used for heart rhythm detection, then the system structure is simple, but the detection accuracy is insufficient due to the complexity of medical signals
Solution Approach 1:
The patent replaces traditional algorithmic processing methods with a deep learning-based neural network system. The neural network automatically learns optimal signal processing algorithms from training data, substituting manual algorithm design with automated machine learning approaches. This enables the system to achieve high detection accuracy by learning complex patterns in radar signals that traditional algorithms cannot capture, while the automated learning process handles the complexity internally rather than requiring explicit complex system design.
Solution Approach 2:
The patent transforms the raw radar signal into a two-dimensional spectrogram representation using Short-Time Fourier Transform (STFT). This parameter transformation converts the one-dimensional time-series signal into a two-dimensional time-frequency representation, making the heartbeat patterns more distinguishable and enabling the neural network to effectively learn from the transformed data with higher accuracy.
2Reliability
If more signal processing and analysis methods are added to solve uncertainty issues, then the detection reliability improves, but the system complexity increases
Solution Approach 1:
The patent employs a neural network that automatically adapts to different detection scenarios through training on diverse datasets. The system performs self-service by automatically learning to handle various uncertainty conditions (different body types, environmental interferences, sensor positions) without requiring manual configuration of multiple specialized processing algorithms. The neural network internally adjusts its parameters and learned features to optimize performance across different conditions, reducing the need for external complex processing chains.
Solution Approach 2:
The patent creates a universal deep learning model that can handle multiple detection challenges simultaneously. A single neural network architecture is trained to process various types of radar signals under different conditions (respiratory harmonics, clutter interference, sensor shaking, unknown body conditions), making the system multi-functional. This universal model replaces the need for separate specialized processing algorithms for each type of uncertainty, achieving high reliability through a consolidated system rather than multiple separate processing chains.
3Measurement precision
If deep learning algorithms are used to automatically extract heartbeat frequency, then the detection accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent applies Short-Time Fourier Transform (STFT) as a preliminary processing step before feeding the signal into the neural network. This preliminary action transforms the raw signal into a spectrogram that pre-highlights the frequency components, making the heartbeat patterns more prominent and easier for the neural network to detect. By performing this frequency transformation beforehand, the neural network receives pre-processed input that reduces the computational burden during the main detection phase while maintaining high accuracy.
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 effectively addresses uncertainty issues by automatically extracting heartbeat frequencies, improving detection accuracy and reliability in various environments, and enabling more precise heart rhythm analysis.
Implementation Method 1
collecting a raw signal by aligning at least one radar sensor toward at least one subject
Implementation Method 2
converting the collected raw signal to two dimensional image information by means of the Fourier Transform
Implementation Method 3
learning from the obtained two dimensional image information and automatically filtering out the noise signal by means of the neural network model so as to learn the relationship between the heartbeat frequency and non-heartbeat frequency
Data Source
AI summary
A heart rhythm detection method and system by using radar sensor is capable of collecting an original signal using a radar sensor toward at least one subject, and converting the original signal to a two dimensional image information (i.e., spectrogram) using the concept of image vision. Then, the neural network automatically learns which heartbeat frequency should be focused on and which heartbeat frequency should be filtered out in the two dimensional image information through deep learning, so that the heartbeat frequencies can be extracted effectively.