repository
Development meaningful feature set for electrocardiogram waveform analysis using unsupervised deep learning algorithms
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.advisor | 윤덕용 | - |
| dc.contributor.author | 장종환 | - |
| dc.date.accessioned | 2022-11-29T02:32:42Z | - |
| dc.date.available | 2022-11-29T02:32:42Z | - |
| dc.date.issued | 2021-02 | - |
| dc.identifier.other | 30708 | - |
| dc.identifier.uri | https://dspace.ajou.ac.kr/handle/2018.oak/20275 | - |
| dc.description | 학위논문(박사)--아주대학교 일반대학원 :의학과,2021. 2 | - |
| dc.description.tableofcontents | I. Introduction 1 A. Background 1 1. Electrocardiogram analysis 1 (A) Electrocardiogram 1 (B) Previous ECG analysis methods 2 (C) Deep learning approach for ECG analysis 10 B. Purpose of this study 13 II. Methods 14 A. Data resource 17 1. Ajou university medical center 18 2. Shaoxing hospital 19 3. BIH-MIT 21 B. Model 23 1. Unsupervised learning 23 2. Autoencoder 24 (A) Autoencocder structure in our study 27 3. Variational autoencoder 42 C. Evaluation 45 1. Model evaluation 45 (A) Autoencocder 45 (B) Variational Autoencocder 45 2. Evaluation of features 46 (A) Anomaly detection 47 (B) Clustering 49 (C) Transfer learning 53 (D) Latent space exploration 58 III. Results 59 A. Data resource 59 B. Autoencoder 61 1. Reconstruction performance 61 2. Clustering 62 3. Transfer learning 65 C. Variational autoencoder 70 1. Reconstruction performance 70 2. Clustering 72 3. Transfer learning 75 4. Anomaly detection 82 5. Latent space exploration 85 IV Discussion 87 V. Conclusion 93 VI. References 94 | - |
| dc.language.iso | eng | - |
| dc.publisher | The Graduate School, Ajou University | - |
| dc.rights | 아주대학교 논문은 저작권에 의해 보호받습니다. | - |
| dc.title | Development meaningful feature set for electrocardiogram waveform analysis using unsupervised deep learning algorithms | - |
| dc.type | Thesis | - |
| dc.contributor.affiliation | 아주대학교 일반대학원 | - |
| dc.contributor.department | 일반대학원 의학과 | - |
| dc.date.awarded | 2021. 2 | - |
| dc.description.degree | Doctoral | - |
| dc.identifier.localId | 1218611 | - |
| dc.identifier.uci | I804:41038-000000030708 | - |
| dc.identifier.url | http://dcoll.ajou.ac.kr:9080/dcollection/common/orgView/000000030708 | - |
| dc.subject.keyword | Autoencoder | - |
| dc.subject.keyword | Deep learning | - |
| dc.subject.keyword | Electrocardiogram (ECG) | - |
| dc.subject.keyword | feature extraction | - |
| dc.subject.keyword | unsupervised learning | - |
| dc.description.alternativeAbstract | Most existing electrocardiogram (ECG) feature extraction methods rely on rule-based approaches. It is difficult to manually define all ECG features. We propose an unsupervised feature learning method using an autoencoder and variational autoencoder (VAE) that can extract ECG features with unlabeled data. Autoencoder was trained using over 2,000,000 ECG samples from 26,481 patients and VAE was trained using 596,000 ECG samples from 1,278 patients, respectively. Two external datasets, which were, Shaoxing and MIT-BIH dataset, were used for feature validation using two approaches. First, we explored the features without an additional training process. Clustering, latent space exploration, and anomaly detection were conducted. We confirmed that ECG features from models reflected the various types of ECG rhythms. Second, we applied ECG features to new tasks as input data and model’s encoder weights to weight initialization for different models as transfer learning for the classification of 12 types of arrhythmias. For evaluation of autoencoder, features, transfer learning and clustering were applied. For evaluation of VAE, two more methods, which are anomaly detection and latent space exploration, were applied including transfer learning and clustering. In experiments of transfer learning using features from unsupervised model, the performance of arrhythmia classification was improved when weight initialization was applied. The f1-score for arrhythmia classification with XGBoost were 0.85, 0.86 using autoencoder and VAE features only, respectively. We confirmed that features from models can be clustered reflecting its characteristics. Moreover, in additional experiments for VAE, we found that its features implied anomality of ECG and its feature space imply clinical meaning for ECG. We confirmed that unsupervised feature learning can extract the characteristics of various types of ECGs and can be an alternative to the feature extraction method for ECGs. | - |
- Appears in Collections:
- Graduate School of Ajou University > Department of Medicine > 4. Theses(Ph.D)
- Files in This Item:
- There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.
-
- License
- STATISTICS
- Total Visit :3,350,603
- Total Download :1,746
- Today View :4,764
