Digitizing Printed ECG Signals Using Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for digitizing printed electrocardiogram (ECG) signals introduce additional noise and are tedious and time-consuming, particularly when converting printed records into digital formats for analysis.
Innovation Solution
The use of machine learning algorithms, specifically a system involving multiple neural networks (division, segmentation, and extraction neural networks) to detect layout regions, segment areas of interest, and extract coordinates from printed ECG images, minimizing noise introduction and automating the digitization process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If histogram filtering, vertical scanning, thresholding and median filtering are used to retrieve digital signal from scanned image, then the digitization process can be completed, but additional noise is introduced on the digitized image
Solution Approach 1:
The patent segments the ECG image processing into distinct functional modules: a layout analysis module that identifies the overall structure and grid lines, a signal extraction module that isolates the actual ECG waveform from background elements, and a coordinate extraction module that converts visual positions to digital coordinates. This segmentation allows each module to be optimized independently, reducing noise introduction while maintaining automation.
Solution Approach 2:
The patent introduces intermediary processing steps between image scanning and final digitization, including preprocessing filters that remove common artifacts before main processing, and post-processing validation that detects and corrects noise-introduced errors. These intermediary layers act as buffers that protect the final output from noise while preserving the automated workflow.
2Loss of information
If manual analysis and computation of printed ECG records is performed, then data can be recovered, but the process is tedious and time consuming
Solution Approach 1:
The patent replaces manual mechanical analysis processes with automated computational systems. The layout analysis module automatically detects grid lines, lead positions, and signal boundaries without human intervention. The signal extraction module uses algorithmic approaches to isolate ECG waveforms from printed records, eliminating the need for manual tracing and measurement. This substitution dramatically reduces both time loss and information loss simultaneously.
Solution Approach 2:
The system performs self-service by automatically adapting to different ECG formats and layouts without requiring manual configuration. The layout analysis module learns from the input image structure and automatically adjusts processing parameters. The coordinate extraction module self-calibrates based on detected grid patterns, enabling rapid processing of diverse printed records without human intervention or time consumption.
Data Source
AI summary
The disclosure relates to systems and methods of converting a representation of a physiological signal (e.g., a non-digitized version such as a printed curve) into a digitized representation of the physiological signal of a subject. For example, a printed electrocardiogram (ECG) may be digitized using the systems are methods provided herein. The method may include receiving a digitized image of a printed curve representing the physiological signal of the subject, and detecting at least one region of interest having a portion of the physiological signal. For each of the regions of interest, the method may include extracting coordinates representing the physiological signal and registering the extracted coordinates.