TY - GEN

T1 - ECG printout interpretation system for clinical decision support

AU - Snegireva, Ekaterina

AU - Khazankin, Grigory R.

AU - Mikheenko, Igor

PY - 2020/7

Y1 - 2020/7

N2 - Nowadays, analog electrocardiographs that deliver only paper printout are ubiquitous in medical institutions. Doctors do visual analysis of electrocardiograms (ECG), occasionally using measurement tools. This article reviews approaches to automatic analysis of electrocardiogram images, including the signal conversion from paper to digital format. The following methods are presented: digitizing graphs from images, determination of signal nodes, and preparation of final report. Various methods of computer vision were tested on electrocardiogram images in order to highlight the graph and transfer coordinates to millimeters. Their limitations are identified and described. Based on the evaluation, a suitable electrocardiogram analysis method has been developed. It includes color filtering of the background grid. Methods of signal analysis and reading of indicators, and their further analysis, are also given. The text conclusion is based on decision trees traversal. As a result, the architecture of measuring system software for electrocardiogram analysis was developed. The system is described considering that the electrocardiogram evaluation unit does not depend on external implementation and can be reused in other systems performing electrocardiogram analysis.

AB - Nowadays, analog electrocardiographs that deliver only paper printout are ubiquitous in medical institutions. Doctors do visual analysis of electrocardiograms (ECG), occasionally using measurement tools. This article reviews approaches to automatic analysis of electrocardiogram images, including the signal conversion from paper to digital format. The following methods are presented: digitizing graphs from images, determination of signal nodes, and preparation of final report. Various methods of computer vision were tested on electrocardiogram images in order to highlight the graph and transfer coordinates to millimeters. Their limitations are identified and described. Based on the evaluation, a suitable electrocardiogram analysis method has been developed. It includes color filtering of the background grid. Methods of signal analysis and reading of indicators, and their further analysis, are also given. The text conclusion is based on decision trees traversal. As a result, the architecture of measuring system software for electrocardiogram analysis was developed. The system is described considering that the electrocardiogram evaluation unit does not depend on external implementation and can be reused in other systems performing electrocardiogram analysis.

KW - biomedical signals analysis

KW - computer vision

KW - electrocardiogram

KW - electrocardiography

KW - graphic filters

KW - image filtering

KW - image processing theory

KW - measuring system

KW - pattern recognition

UR - http://www.scopus.com/inward/record.url?scp=85094808740&partnerID=8YFLogxK

UR - https://www.elibrary.ru/item.asp?id=45184403

U2 - 10.1109/CSGB51356.2020.9214740

DO - 10.1109/CSGB51356.2020.9214740

M3 - Conference contribution

AN - SCOPUS:85094808740

T3 - Proceedings - 2020 Cognitive Sciences, Genomics and Bioinformatics, CSGB 2020

SP - 19

EP - 22

BT - Proceedings - 2020 Cognitive Sciences, Genomics and Bioinformatics, CSGB 2020

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2020 Cognitive Sciences, Genomics and Bioinformatics, CSGB 2020

Y2 - 6 July 2020 through 10 July 2020

ER -