Clinical electrocardiogram digitization based on U²Net

Abstract

Aim: Deep learning has advanced medical artificial intelligence, but many electrocardiogram (ECG) records remain available only as paper-based documents. Scanned paper ECG images often contain grid lines, background artifacts, and unstructured waveform information, which limits their direct use for quantitative ECG analysis. This study aimed to develop an improved ECG image segmentation method for extracting clean ECG traces from scanned paper-based ECG records and supporting subsequent ECG signal digitization.

Methods: We proposed CU²-NeXt, an ECG image segmentation framework that extends U²-Net with ConvNeXt modules and incorporates convolutional block attention module (CBAM) to emphasize waveform-related features while suppressing background artifacts. Viterbi dynamic programming was subsequently applied to reconstruct continuous signal trajectories from the segmented waveform pixels.

Results: On the evaluated dataset, CU²-NeXt achieved a macro-F1 score of 88.9% and a Dice coefficient of 78.7%, with higher overlap-based performance than the evaluated baseline segmentation models.

Conclusion: These findings demonstrate the feasibility of using CU²-NeXt for ECG trace extraction and digitization from scanned paper ECG images. The proposed framework may provide a useful preprocessing approach for converting paper ECG records into analyzable digital signals and supporting downstream quantitative ECG analysis.

Keywords

Paper ECG / ECG image segmentation / viterbi algorithm / deep learning / grid removal

References

[1] Roth GA,Abate KH.Global, regional, and national age-sex-specific mortality for 282 causes of death in 195 countries and territories, 1980-2017: a systematic analysis for the global burden of disease study 2017.Lancet2018;392:1736-88

[2] di Bernardo D.Explaining the T-wave shape in the ECG.Nature2000;403:40

[3] Attia ZI,Dugan J.Prospective evaluation of smartwatch-enabled detection of left ventricular dysfunction.Nat Med2022;28:2497-503 PMCID:PMC9805528

[4] Holst H,Peterson C.A confident decision support system for interpreting electrocardiograms.Clin Physiol1999;19:410-8

[5] Pan J.A real-time QRS detection algorithm.IEEE Trans Biomed Eng1985;32:230-6

[6] Belforte G,Ferraris F.A contribution to the automatic processing of electrocardiograms using syntactic methods.IEEE Trans Biomed Eng1979;26:125-36

[7] Raghunath S,Jing L.Prediction of mortality from 12-lead electrocardiogram voltage data using a deep neural network.Nat Med2020;26:886-91

[8] Lence A,Fall A,Zucker JD.Automatic digitization of paper electrocardiograms - a systematic review.J Electrocardiol2023;80:125-32

[9] Poterucha TJ,Ricart RP.Detecting structural heart disease from electrocardiograms using AI.Nature2025;644:221-30

[10] Al-Zaiti SS,Zègre-Hemsey JK.Machine learning for ECG diagnosis and risk stratification of occlusion myocardial infarction.Nat Med2023;29:1804-13. PMCID:PMC10353937

[11] Attia ZI,Behr ER.Application of artificial intelligence to the electrocardiogram.Eur Heart J2021;42:4717-30 PMCID:PMC8500024

[12] Siontis KC,Attia ZI.Artificial intelligence-enhanced electrocardiography in cardiovascular disease management.Nat Rev Cardiol2021;18:465-78 PMCID:PMC7848866

[13] Kligfield P,Brown B,Kohls M.The ISCE ECG genome pilot challenge: a 2004 progress report.J Electrocardiol2004;37 Suppl:144-8

[14] Stockbridge N.Points to consider in electrocardiogram waveform extraction.J Electrocardiol2005;38:319-20

[15] Badilini F,Zareba W.ECGScan: a method for conversion of paper electrocardiographic printouts to digital electrocardiographic files.J Electrocardiol2005;38:310-8

[16] Wu H,Li X.A fully-automated paper ECG digitisation algorithm using deep learning.Sci Rep2022;12:20963 PMCID:PMC9722713

[17] Randazzo V,Paviglianiti A,Pasero E.Development and validation of an algorithm for the digitization of ECG paper images.Sensors2022;22:7138 PMCID:PMC9572306

[18] Wang T,Chang ZS,Huang ZP.A fully-automated paper ECG digitisation algorithm for e-health applications. In: 2024 16th International Conference on Wireless Communications and Signal Processing (WCSP); 2024 Oct 24-26; Hefei, China. IEEE; 2024. pp. 1295-300.

[19] Ganesh S,Alkhalaf M,Shah AJ.Combining optical character recognition with paper ECG digitization.IEEE J Transl Eng Health Med2021;9:1900309 PMCID:PMC8248903

[20] Li Y,Wang M.Deep learning for digitizing highly noisy paper-based ECG records.Comput Biol Med2020;127:104077

[21] Patil R,Varma S,Mehendale N.Auto lead extraction and digitization of ECG paper records using cGAN.ArXiv2022;arXiv:2211.06720

[22] Shivashankara KK,Shervedani AM,Reyna MA.ECG-Image-Kit: a synthetic image generation toolbox to facilitate deep learning-based electrocardiogram digitization.Physiol Meas2024;45:055019 PMCID:PMC11135178

[23] Reyna MA,Weigle J.Ecg-image-database: a dataset of ecg images with real-world imaging and scanning artifacts; a foundation for computerized ecg image digitization and analysis.ArXiv2024;arXiv:2409.16612

[24] Fortune JD,Haq KT,Tereshchenko LG.Digitizing ECG image: a new method and open-source software code.Comput Methods Programs Biomed2022;221:106890 PMCID:PMC9286778

[25] Qin X,Huang C,Zaiane OR.U2-Net: going deeper with nested U-structure for salient object detection.Pattern Recognition2020;106:107404

[26] Liu Z,Wu CY,Darrell T.A ConvNet for the 2020s. In: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR); 2022 Jun 18-24; New Orleans, LA, USA. IEEE; 2022. pp. 11966-76.

[27] Woo S,Lee JY.CBAM: convolutional block attention module. In: Ferrari V, Hebert M, Sminchisescu C, Weiss Y, Editors. Computer Vision - ECCV 2018: 15th European Conference; 2018 Sep 8-14; Munich, Germany. Springer, Cham; 2018. pp. 3-19.

[28] Lin TY,Girshick R,Dollár P.Focal loss for dense object detection. In: 2017 IEEE International Conference on Computer Vision (ICCV); 2017 Oct 22-29; Venice, Italy. IEEE; 2017. pp. 2999-3007.

[29] Milletari F,Ahmadi SA.V-net: fully convolutional neural networks for volumetric medical image segmentation. In: 2016 fourth international conference on 3D vision (3DV); 2016 Oct 25-28; Stanford, CA, USA. IEEE; 2016. pp. 565-71.

[30] International Electrotechnical Commission. IEC 60601-2-25:2011. Medical electrical equipment - Part 2-25: Particular requirements for the basic safety and essential performance of electrocardiographs. 2nd ed. Geneva: International Electrotechnical Commission; 2011. Available from: https://webstore.iec.ch/en/publication/2636 [accessed 18 July 2026]

[31] Baydoun M,Abou Hassan OK,El Hajj A.High precision digitization of paper-based ECG records: a step toward machine learning.IEEE J Transl Eng Health Med2019;7:1900808 PMCID:PMC6876931