Combining Optical Character Recognition With Paper ECG Digitization

Abstract

Objective: We propose a MATLAB-based tool to convert electrocardiography (ECG) waveforms from paper-based ECG records into digitized ECG signals that is vendor-agnostic. The tool is packaged as an open source standalone graphical user interface (GUI) based application. Methods and procedures: To reach this objective we: (1) preprocess the ECG records, which includes skew correction, background grid removal and linear filtering; (2) segment ECG signals using Connected Components Analysis (CCA); (3) implement Optical Character Recognition (OCR) for removal of overlapping ECG lead characters and for interfacing of patients’ demographic information with their research records or their electronic medical record (EMR). The ECG digitization results are validated through a reader study where clinically salient features, such as intervals of QRST complex, between the paper ECG records and the digitized ECG records are compared. Results: Comparison of clinically important features between the paper-based ECG records and the digitized ECG signals, reveals intra- and inter-observer correlations of 0.86–0.99 and 0.79–0.94, respectively. The kappa statistic was found to average at 0.86 and 0.72 for intra- and inter-observer correlations, respectively. Conclusion: The clinically salient features of the ECG waveforms such as the intervals of QRST complex, are preserved during the digitization procedure. Clinical and Healthcare Impact: This open-source digitization tool can be used as a research resource to digitize paper ECG records thereby enabling development of new prediction algorithms to risk stratify individuals with cardiovascular disease, and/or allow for development of ECG-based cardiovascular diagnoses relying upon automated digital algorithms.

ECG-Image-Kit: a synthetic image generation toolbox to facilitate deep learning-based electrocardiogram digitization

“…In the ECG context, various digitization methods have been proposed; including the grayscale thresholding and contour-based digitization method by Ravichandran et al ( 2013 ), color segmentation and median filtering for noise removal by Garg et al ( 2012 ), and the combination of optical character recognition (OCR) with image processing techniques for digitization and artifact removal (Baydoun et al 2019 , Ganesh et al 2021 ). However, classical image processing methods, sensitive to input quality and environmental artifacts, often struggle with low-quality, distorted paper ECG records.…”

“…The user of the toolbox has the option to choose whether there should be an overlap between ECG segments and printed text artifacts. Although overlapped characters pose a problem in digitizing paper ECG records (Ganesh et al 2021 ), they are added to represent realistic paper ECGs, which occasionally print text with partial overlap with the ECG traces. Further, to add other printed information such as date, patient record numbers, etc the toolkit uses the corresponding fields from the WFDB header files that accompany all PhysioNet data files, or through a customizable text-based template file.…”

Research on end-to-end document image character recognition based on differentiable binarization

“…The image character recognition model is the second step in the two steps of detection and recognition. The document character recognition model is also simpler than the character recognition task in the natural scene [5] . Since there is a large difference between the foreground and background in the foreground distribution map, a relatively simple global threshold is selected here M calculate the average value of all pixels in the foreground distribution according to the formula P and standard deviation λ is the available feature extraction algorithm is:…”

Invoice Recognition System based on Neural Network

“…Due to the shooting light, shooting background, shooting Angle and the lack of cleanliness of the invoice itself, the invoice image will have strong noise and even the information on the invoice cannot be recognized [11]. Direct use of the original invoice image to identify, will cause a great error in the identification results, resulting in poor identification results.…”