Printed ECG Image Analysis for Accessible Cardiovascular Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cardiovascular disease detection methods relying on electrocardiographic signal-based models are limited by their reliance on raw data, lack of generalizability across institutions, and lack of interpretability, making them inaccessible for widespread use in family practice clinics and remote settings.
Innovation Solution
A computer-implemented method using a machine-learning algorithm, such as a deep neural network, to analyze printed electrocardiogram images, capable of distinguishing healthy and diseased heart readings, and adaptable to various formats and noise characteristics, allowing for decentralized detection of cardiovascular diseases.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If signal-based AI algorithms are used for cardiovascular disease detection, then detection accuracy is improved, but accessibility and ease of operation deteriorate due to reliance on raw data and single-source training
Solution Approach 1:
The patent uses printed ECG waveform images as copies of the original electrocardiographic data. Instead of requiring access to raw signal data from specific ECG machines, the system processes printed images that can be obtained from any ECG device. This copying approach maintains detection accuracy while significantly improving accessibility, as printed images are universally available across different healthcare settings and can be easily shared and stored.
Solution Approach 2:
The deep learning model is trained to process ECG images from multiple sources and formats, making it universal rather than specific to single-source data. The system can handle images from different ECG machines, with varying resolutions, orientations, and quality levels. This multi-functionality ensures both high detection accuracy and broad accessibility across diverse healthcare environments.
2Ease of manufacture
If models are trained on data from a single source, then training simplicity is improved, but generalizability to different institutions and health settings deteriorates
Solution Approach 1:
The training process incorporates parameter changes by exposing the model to ECG images with varying characteristics including different resolutions, orientations, noise levels, and formatting styles. By deliberately introducing these parameter variations during training, the model learns robust features that generalize well across different institutions and health settings, overcoming the limitations of single-source training while maintaining training feasibility.
3Measurement precision
If advanced diagnostic tools such as echocardiography, CT, or MRI are used for screening, then detection capability is improved, but cost and barriers to use increase
Solution Approach 1:
The patent replaces complex mechanical diagnostic systems (echocardiography machines, CT scanners, MRI devices) with a computational system that processes simple printed ECG images. This substitution maintains detection capability for cardiovascular diseases while dramatically reducing device complexity and cost barriers, enabling screening in resource-limited settings and primary care environments where advanced imaging equipment is unavailable.
Data Source
AI summary
Provided herein are computer-implemented methods of detecting cardiovascular disease in a subject. The methods include receiving an electrocardiogram (ECG) image for the subject; applying a machine-learning based algorithm to the ECG image for the subject, the algorithm being trained to distinguish a printed ECG reading of a heart with cardiovascular disease from a printed ECG reading of a healthy heart; comparing outputs of the algorithm to patterns of algorithm outputs for ECG images from healthy subjects and subjects with one or more cardiovascular diseases; and determining if the subject has cardiovascular disease based upon the outputs of the algorithm.