kth.sePublications KTH
Operational message
There are currently operational disruptions. Troubleshooting is in progress.
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
ECG Analysis Using Deep Learning: Exploring Data Augmentations' Effect on Model Performance
KTH, School of Electrical Engineering and Computer Science (EECS).
KTH, School of Electrical Engineering and Computer Science (EECS).
2025 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

The electrocardiogram (ECG) is a crucial tool for identifying heart abnormalities in a non-invasive way. As the technology has advanced, it is now a digital tool that can integrate algorithms to support abnormality diagnostics. Deep neural networks (DNN) have, in the past few years, shown their potential and success in classification problems, and studies have shown that a DNN can outperform a cardiologist in diagnosing abnormalities from ECG. In this study, we investigate whether data augmentation (DA) can improve model performance. Using a DNN model trained on PTB-XL, a large publicly available electrocardiography dataset, we aim to apply different augmentations and evaluate DAs effect on model performance using the F1 metric. The results show that combinations of carefully selected DA can improve model accuracy with a significant result.

Abstract [sv]

Elektrokardiogram (EKG) är ett viktigt verktyg för att på ett icke-invasivt sätt upptäcka hjärtproblem. Med den tekniska utvecklingen har EKG blivit digitalt och man kan därmed integrera olika algoritmer som stödjer diagnostiken av avvikelserna. Under de senaste åren har djupa neurala nätverk (DNN) visat stor potential och framgång inom klassificeringsproblem och forskning har visat att DNN-modeller kan överträffa kardiologer i att identifiera avvikelser i EKG. I denna studie undersöker vi om dataaugmentering (DA) kan förbättra en DNN-modells prestanda. Vi använder en DNN-modell som tränats på PTB-XL, a large publicly available electrocardiography dataset, en av de största datasets som finns för 12-avlednings-EKG. Vi tillämpar olika typer av dataaugmenteringar och utvärderar sedan effekterna på modellens prestanda med hjälp av F1-måttet. Resultaten visar att noggrant utvalda kombinationer av dataaugmentering kan ge en signifikant förbättring av modellens prestanda.

Place, publisher, year, edition, pages
2025. , p. 499-504
Series
TRITA-EECS-EX ; 2025:149
National Category
Electrical Engineering, Electronic Engineering, Information Engineering
Identifiers
URN: urn:nbn:se:kth:diva-376171OAI: oai:DiVA.org:kth-376171DiVA, id: diva2:2034539
Supervisors
Examiners
Projects
Kandidatexamensarbete i Elektroteknik 2025, EECS, KTHAvailable from: 2026-02-02 Created: 2026-02-02

Open Access in DiVA

fulltext(80627 kB)15 downloads
File information
File name FULLTEXT01.pdfFile size 80627 kBChecksum SHA-512
35ce0a386dafe4649eb99cbe0efdfed651a3c9044e3339612422234d17a7e8ec21d4fd4aa201500c3c7a8f57194994b78b3e0cfbd5319ecd49f18a5d8a7ff775
Type fulltextMimetype application/pdf

By organisation
School of Electrical Engineering and Computer Science (EECS)
Electrical Engineering, Electronic Engineering, Information Engineering

Search outside of DiVA

GoogleGoogle Scholar
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 3957 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf