Fetal ECG Classification Using VCG and Motion Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current prenatal detection methods for congenital heart disease (CHD) using ultrasonic examination have low detection rates and lack easy guidelines for interpreting fetal electrocardiograms (fECGs), making it difficult to identify abnormal ECGs associated with CHD, especially before 20 weeks of gestation.
Innovation Solution
A device and method for classifying fetal electrocardiograms using machine learning classifiers, which process and filter fECG signals to generate vector cardiograms (VCGs), compensate for fetal movement, and normalize data to improve signal quality, enabling accurate classification of normal or abnormal fetal heart conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ultrasonic examination is used for prenatal detection of CHD, then screening can be performed, but detection rates remain low and interpretation is difficult
Solution Approach 1:
The patent introduces a machine learning classifier as an intermediary between the raw fetal ECG signals and the diagnostic decision. This classifier automatically processes the complex ECG data, extracting meaningful patterns and providing clear classification results, thereby improving both detection accuracy and ease of interpretation without requiring expert manual analysis
Solution Approach 2:
The patent replaces the manual mechanical interpretation process of fetal ECG signals by clinicians with an automated machine learning system. The classifier automatically analyzes ECG features, compensates for fetal movements, and generates diagnostic classifications, substituting human interpretation effort with computational processing that provides more consistent and accurate results
2Measurement precision
If machine learning classifier is used to classify fetal ECG, then CHD detection accuracy improves, but device complexity increases
Solution Approach 1:
The patent segments the complex classification task into distinct modular processing stages: signal acquisition, noise filtering, fetal movement compensation, feature extraction, and final classification. Each stage handles a specific aspect of the problem independently, making the overall complex system more manageable and interpretable while maintaining high accuracy
Solution Approach 2:
The patent performs preliminary processing actions on the fetal ECG signals before classification, including filtering out maternal ECG interference, compensating for fetal movements, and extracting relevant features. This preliminary preparation simplifies the subsequent classification task and improves accuracy without requiring the classifier to handle raw, unprocessed signals
3Loss of time
If fetal ECG signals are processed early in gestation, then earlier CHD detection is possible, but signal quality is poorer
Solution Approach 1:
The patent converts the harmful effect of fetal movements, which increase with gestation age and degrade signal quality, into a beneficial feature. By using the characteristic patterns of fetal movements in the ECG signals, the system compensates for and corrects the signal distortion caused by these movements, thereby maintaining high classification accuracy even when processing signals from later gestation periods
Solution Approach 2:
The patent dynamically adjusts processing parameters based on gestation age and signal characteristics. The machine learning classifier is trained to handle varying signal qualities across different gestation periods, adapting its processing approach to optimize detection accuracy for each developmental stage rather than using a fixed processing pipeline
Data Source
Figure 1a~2a
Figure 2b
Figure 3~4
AI summary
Some embodiments are directed to a device (200) for processing a fetal electrocardiogram (fECG). The processing may include obtaining a fetal vector cardiogram (VCG) from said multiple ECG signals, and providing the fetal vector cardiogram as an input to the machine learning classifier configured with trained parameters, and obtain a classification from the machine learning classifier.