Fetal ECG Extraction Using Reference Signal ICA
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for extracting fetal electrocardiogram signals from a pregnant mother's abdominal electrocardiogram signals face challenges such as distinguishing fetal signals from maternal and electromyogram noise, difficulty in extracting P-waves and T-waves, unpredictable lead type, and reduced precision due to high-amplitude maternal signals, especially during 26th to 36th weeks of pregnancy.
Innovation Solution
An electrocardiogram signal-processing method using a reference signal generated from the fetal heartbeat period, combined with blind source separation and independent component analysis, to actively extract fetal electrocardiogram signals of a designated lead type from biopotential signals detected on the mother's abdomen, while removing maternal electrocardiogram signals and optimizing electrode placement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If independent component analysis is used to extract fetal electrocardiogram signals, then signal extraction can be performed, but the extraction precision deteriorates due to contamination by electromyogram signals and other signals with similar probability distributions
Solution Approach 1:
The patent introduces an electromyogram signal as an intermediary reference signal in the independent component analysis process. By using the detected electromyogram signal as a reference, the algorithm can identify and separate components correlated with maternal muscle activity, thereby reducing contamination in the extracted fetal electrocardiogram signal and improving extraction precision
Solution Approach 2:
The patent modifies the independent component analysis by incorporating additional parameters: the detected electromyogram signal and its time derivative are used as reference signals. This changes the analysis parameters from standard ICA to a constrained version that uses external physiological references, enabling better separation of overlapping signals
2Object-affected harmful factors
If abdominal electrodes are used for non-invasive measurement, then patient safety is improved, but the signal-to-noise ratio deteriorates due to high-amplitude maternal signals and fetal signals during 26th to 36th weeks of pregnancy
Solution Approach 1:
The patent uses detected electromyogram signals as intermediary references to mediate the separation process between maternal and fetal signals. By incorporating these intermediate physiological measurements into the analysis, the system can better distinguish fetal signals from maternal interference without requiring invasive procedures
Solution Approach 2:
The patent performs preliminary detection and processing of electromyogram signals before using them as references in the independent component analysis. This preliminary action of characterizing maternal muscle activity enables the subsequent extraction algorithm to preemptively compensate for electromyogram contamination in the fetal signal extraction
3Measurement precision
If scalp electrodes are attached to the fetus during delivery for direct measurement, then signal quality is improved, but patient safety deteriorates due to invasive procedures and increased infection risk
Solution Approach 1:
The patent creates a virtual reference signal by detecting and processing electromyogram activity from abdominal electrodes. This copied physiological signal serves as a surrogate for direct fetal signal measurement, allowing the system to achieve good signal quality through computational separation rather than through invasive physical contact with the fetus
4Device complexity
If conventional extraction methods are used, then processing simplicity is maintained, but the ability to extract P-waves and T-waves deteriorates due to difficulty in distinguishing fetal signals from maternal and noise signals
Solution Approach 1:
The patent introduces detected electromyogram signals as intermediary references that enable the independent component analysis to separately identify P-waves, T-waves, and other fetal electrocardiogram components. These intermediary signals provide additional constraints that help resolve the ambiguity in distinguishing fetal waves from maternal and noise signals
Solution Approach 2:
The patent uses the time derivative of the detected electromyogram signal as a dynamic reference that adapts to changing physiological conditions. This dynamic reference signal allows the extraction algorithm to maintain capability for extracting P-waves and T-waves across different gestational ages and fetal states
Data Source
AI summary
An electrocardiogram signal processing method and device for extracting a fetus electrocardiogram signal included in a biopotential signal detected from an electrode attached to a mother's body is provided. The electrocardiogram signal processing method for extracting an electrocardiogram signal of a fetus (1b) from a biopotential signal containing electrocardiogram signals of the mother's body (1) and a fetus (1b) inputted through an electrode (E) attached to the mother's body (1) during pregnancy, comprises a reference signal generating step 4 of generating a reference signal for separating/extracting a fetus electrocardiogram signal of a specified induction type according to a beat rate signal inputted from a sensor (S) for detecting the beat rate of the fetus (1b) and a fetus electrocardiogram signal extracting step 5 of separating/extracting the fetus electrocardiogram signal of the specified induction type from the biopotential signal inputted through an electrode (E) according to the reference signal generated at reference signal generating step 4 by a reference system independent component analyzing method.