Analisis Teknik Explainable AI Post-hoc untuk Deteksi Infark Miokard Berbasis Sinyal EKG
Abstract
Model deep learning (DL) menunjukkan akurasi yang unggul dalam menganalisis sinyal EKG untuk mendeteksi infark miokard. Namun, pendekatan ini sering dianggap sebagai "black box" yang sulit dipahami oleh tenaga medis. Penelitian ini mengevaluasi penerapan explainable AI (XAI), yakni Grad-CAM dan LRP, dalam meningkatkan interpretabilitas model DL berbeda untuk deteksi infark miokard menggunakan sinyal EKG. Pada studi ini, XAI berupa Grad-CAM dan LRP diaplikasikan pada tiga jenis model konvolusional, yakni model konvolusional biasa, model residual, serta model inception. Hasil menunjukkan bahwa Grad-CAM memberikan penjelasan spasial yang konsisten dengan atribusi positif serta lebih sederhana, sementara LRP dapat memberikan atribusi positif maupun negatif, membedakan relevansi antar-lead, serta tidak tergantung pada resolusi spasial dari layer internal model. Kemudian, kombinasi Grad-CAM untuk analisis temporal dan LRP untuk analisis relevansi lead memberikan interpretasi model yang paling komprehensif dan direkomendasikan untuk evaluasi relevansi klinis model DL. Disimpulkan bahwa arsitektur InceptionTime ditemukan merupakan model terbaik, dengan akurasi tertinggi serta utilisasi lead tertinggi berdasarkan analisis XAI.
Kata kunci—deep learning, elektrokardiogram, infark miokard, XAI
References
E. Susanti, “Kemenkes : Penyakit Kardiovaskular Penyebab Kematian Tertinggi di Indonesia,” Dinas Kesehatan Provinsi Aceh. Diakses: 5 November 2024. [Daring]. Tersedia pada: https://dinkes.acehprov.go.id/detailpost/kemenkes-penyakit-kardiovaskular-penyebab-kematian-tertinggi-di-indonesia
P. L. Laforgia, C. Auguadro, S. Bronzato, dan A. Durante, “The Reduction of Mortality in Acute Myocardial Infarction,” Int J Prev Med, vol. 13, no. 1, hlm. 56, Jan 2022, doi: 10.4103/ijpvm.IJPVM_122_20.
E. Radwa, H. Ridha, dan B. Faycal, “Deep learning-based approaches for myocardial infarction detection: A comprehensive review recent advances and emerging challenges,” 1 September 2024, Elsevier B.V. doi: 10.1016/j.medntd.2024.100322.
O. J. Mechanic, M. Gavin, dan S. A. Grossman, Acute Myocardial Infarction. 2024. [Daring]. Tersedia pada: http://www.ncbi.nlm.nih.gov/pubmed/30269080
M. Maleki, “Evaluation of Patient With Cardiovascular Problem,” dalam Practical Cardiology, Elsevier, 2022, hlm. 7–16. doi: 10.1016/B978-0-323-80915-3.00015-6.
S. Ghaffari, “Non-ST-Elevation Acute Coronary Syndromes,” dalam Practical Cardiology, Elsevier, 2022, 23, hlm. 413–428. doi: 10.1016/B978-0-323-80915-3.00028-4.
J. Hassannataj Joloudari dkk., “Application of artificial intelligence techniques for automated detection of myocardial infarction: a review,” 31 Agustus 2022, Institute of Physics. doi: 10.1088/1361-6579/ac7fd9.
L. Karatzia, N. Aung, dan D. Aksentijevic, “Artificial intelligence in cardiology: Hope for the future and power for the present,” 13 Oktober 2022, Frontiers Media S.A. doi: 10.3389/fcvm.2022.945726.
W.-C. Liu dkk., “A deep learning algorithm for detecting acute myocardial infarction,” EuroIntervention, vol. 17, no. 9, hlm. 765–773, Okt 2021, doi: 10.4244/EIJ-D-20-01155.
P. Linardatos, V. Papastefanopoulos, dan S. Kotsiantis, “Explainable ai: A review of machine learning interpretability methods,” 1 Januari 2021, MDPI AG. doi: 10.3390/e23010018.
E. Tjoa dan C. Guan, “A Survey on Explainable Artificial Intelligence (XAI): Toward Medical XAI,” IEEE Trans Neural Netw Learn Syst, vol. 32, no. 11, hlm. 4793–4813, Nov 2021, doi: 10.1109/TNNLS.2020.3027314.
I. Rehman dan A. Rehman, Anatomy, Thorax, Heart. 2024. [Daring]. Tersedia pada: http://www.ncbi.nlm.nih.gov/pubmed/0
F. Kusumoto, ECG Interpretation. Cham: Springer International Publishing, 2020. doi: 10.1007/978-3-030-40341-6.
D. G. Strauss, D. D. Schocken, dan G. S. Wagner, Marriott’s Practical Electrocardiography. Wolters Kluwer, 2021.
M. J. Alemzadeh-Ansari, “Electrocardiography,” dalam Practical Cardiology, Elsevier, 2022, hlm. 17–59. doi: 10.1016/B978-0-323-80915-3.00005-3.
A. L. Goldberger, Z. D. Goldberger, dan A. Shvilkin, Goldberger’s Clinical Electrocardiography: A Simplified Approach, 9 ed. Philadelphia (PA): Elsevier, 2018.
X. Chen dkk., “Acute Myocardial Infarction Detection Using Deep Learning-Enabled Electrocardiograms,” Front Cardiovasc Med, vol. 8, 2021, doi: 10.3389/fcvm.2021.654515.
M. Ganeshkumar, V. Ravi, V. Sowmya, E. A. Gopalakrishnan, dan K. P. Soman, “Explainable Deep Learning-Based Approach for Multilabel Classification of Electrocardiogram,” IEEE Trans Eng Manag, vol. 70, no. 8, hlm. 2787–2799, Agu 2023, doi: 10.1109/TEM.2021.3104751.
V. Jahmunah, E. Y. K. Ng, R. S. Tan, S. L. Oh, dan U. R. Acharya, “Explainable detection of myocardial infarction using deep learning models with Grad-CAM technique on ECG signals,” Comput Biol Med, vol. 146, Jul 2022, doi: 10.1016/j.compbiomed.2022.105550.
P. Xiong, S. M. Y. Lee, dan G. Chan, “Deep Learning for Detecting and Locating Myocardial Infarction by Electrocardiogram: A Literature Review,” 25 Maret 2022, Frontiers Media S.A. doi: 10.3389/fcvm.2022.860032.
W. J. Tompkins, “A Real-Time QRS Detection Algorithm,” 1985.
H. M. Rai dan K. Chatterjee, “Hybrid CNN-LSTM deep learning model and ensemble technique for automatic detection of myocardial infarction using big ECG data,” Applied Intelligence, vol. 52, no. 5, hlm. 5366–5384, Mar 2022, doi: 10.1007/s10489-021-02696-6.
N. Strodthoff, P. Wagner, T. Schaeffter, dan W. Samek, “Deep Learning for ECG Analysis: Benchmarks and Insights from PTB-XL,” IEEE J Biomed Health Inform, vol. 25, no. 5, hlm. 1519–1528, Mei 2021, doi: 10.1109/JBHI.2020.3022989.
H. Ismail Fawaz dkk., “InceptionTime: Finding AlexNet for time series classification,” Data Min Knowl Discov, vol. 34, no. 6, hlm. 1936–1962, Nov 2020, doi: 10.1007/s10618-020-00710-y.
Y. Jin dkk., “A Novel Interpretable Method Based on Dual-Level Attentional Deep Neural Network for Actual Multilabel Arrhythmia Detection,” IEEE Trans Instrum Meas, vol. 71, hlm. 1–11, 2022, doi: 10.1109/TIM.2021.3135330.
J. Yoo, T. J. Jun, dan Y.-H. Kim, “xECGNet: Fine-tuning attention map within convolutional neural network to improve detection and explainability of concurrent cardiac arrhythmias,” Comput Methods Programs Biomed, vol. 208, hlm. 106281, Sep 2021, doi: 10.1016/j.cmpb.2021.106281.
J. M. Kwon dkk., “Artificial intelligence assessment for early detection of heart failure with preserved ejection fraction based on electrocardiographic features,” European Heart Journal - Digital Health, vol. 2, no. 1, hlm. 106–116, Mar 2021, doi: 10.1093/ehjdh/ztaa015.
C. Han, J. Sun, Y. Bian, W. Que, dan L. Shi, “Automated Detection and Localization of Myocardial Infarction With Interpretability Analysis Based on Deep Learning,” IEEE Trans Instrum Meas, vol. 72, hlm. 1–12, 2023, doi: 10.1109/TIM.2023.3258521.
W. Liu, F. Wang, Q. Huang, S. Chang, H. Wang, dan J. He, “MFB-CBRNN: A Hybrid Network for MI Detection Using 12-Lead ECGs,” IEEE J Biomed Health Inform, vol. 24, no. 2, hlm. 503–514, Feb 2020, doi: 10.1109/JBHI.2019.2910082.
D. Hendrycks dan K. Gimpel, “Gaussian Error Linear Units (GELUs),” arXiv: Learning, Jun 2016, [Daring]. Tersedia pada: http://arxiv.org/abs/1606.08415
R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, dan D. Batra, “Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization,” dalam 2017 IEEE International Conference on Computer Vision (ICCV), IEEE, Okt 2017, hlm. 618–626. doi: 10.1109/ICCV.2017.74.
K. He, X. Zhang, S. Ren, dan J. Sun, “Deep Residual Learning for Image Recognition,” dalam Computer Vision and Pattern Recognition (CVPR), 2016, hlm. 770–778. doi: 10.1109/CVPR.2016.90.
C. Szegedy dkk., “Going Deeper with Convolutions,” 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), hlm. 1–9, Jun 2015.
C. Han, Y. Zhou, W. Que, Z. Li, dan L. Shi, “An Overview of Algorithms for Myocardial Infarction Diagnostics Using ECG Signals: Advances and Challenges,” IEEE Trans Instrum Meas, vol. 73, 2024, doi: 10.1109/TIM.2024.3418105.
C. P. Langlotz dkk., “A Roadmap for Foundational Research on Artificial Intelligence in Medical Imaging: From the 2018 NIH/RSNA/ACR/The Academy Workshop,” Radiology, vol. 291, no. 3, hlm. 781–791, Jun 2019, doi: 10.1148/radiol.2019190613.
S. M. Anwar, “AIM and Explainable Methods in Medical Imaging and Diagnostics,” dalam Artificial Intelligence in Medicine, Springer International Publishing, 2022, hlm. 501–510. doi: 10.1007/978-3-030-64573-1_293.
A. M. Salih dkk., “A review of evaluation approaches for explainable AI with applications in cardiology,” Artif Intell Rev, vol. 57, no. 9, Sep 2024, doi: 10.1007/s10462-024-10852-w.
M. T. Ribeiro, S. Singh, dan C. Guestrin, “‘Why should i trust you?’ Explaining the predictions of any classifier,” dalam Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Association for Computing Machinery, Agu 2016, hlm. 1135–1144. doi: 10.1145/2939672.2939778.
S. Lundberg dan S.-I. Lee, “A Unified Approach to Interpreting Model Predictions,” Nov 2017, [Daring]. Tersedia pada: http://arxiv.org/abs/1705.07874
S. Bach, A. Binder, G. Montavon, F. Klauschen, K. R. Müller, dan W. Samek, “On pixel-wise explanations for non-linear classifier decisions by layer-wise relevance propagation,” PLoS One, vol. 10, no. 7, Jul 2015, doi: 10.1371/journal.pone.0130140.
C. Han, J. Sun, Y. Bian, W. Que, dan L. Shi, “Automated Detection and Localization of Myocardial Infarction With Interpretability Analysis Based on Deep Learning,” IEEE Trans Instrum Meas, vol. 72, 2023, doi: 10.1109/TIM.2023.3258521.
P. Singh dan A. Sharma, “Interpretation and Classification of Arrhythmia Using Deep Convolutional Network,” IEEE Trans Instrum Meas, vol. 71, 2022, doi: 10.1109/TIM.2022.3204316.
Z. I. Attia, G. Lerman, dan P. A. Friedman, “Deep neural networks learn by using human-selected electrocardiogram features and novel features,” European Heart Journal - Digital Health, vol. 2, no. 3, hlm. 446–455, Sep 2021, doi: 10.1093/ehjdh/ztab060.
T. A. A. Abdullah, M. S. M. Zahid, W. Ali, dan S. U. Hassan, “B-LIME: An Improvement of LIME for Interpretable Deep Learning Classification of Cardiac Arrhythmia from ECG Signals,” Processes, vol. 11, no. 2, Feb 2023, doi: 10.3390/pr11020595.
A. Karoui, M. Bendahmane, dan N. Zemzemi, “Cardiac Activation Maps Reconstruction: A Comparative Study Between Data-Driven and Physics-Based Methods,” Front Physiol, vol. 12, Agu 2021, doi: 10.3389/fphys.2021.686136.
G. Montavon, A. Binder, S. Lapuschkin, W. Samek, dan K. R. Müller, “Layer-Wise Relevance Propagation: An Overview,” dalam Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11700 LNCS, Springer Verlag, 2019, hlm. 193–209. doi: 10.1007/978-3-030-28954-6_10.
W. Samek, A. Binder, G. Montavon, S. Lapuschkin, dan K.-R. Muller, “Evaluating the Visualization of What a Deep Neural Network Has Learned,” IEEE Trans Neural Netw Learn Syst, vol. 28, no. 11, hlm. 2660–2673, Nov 2017, doi: 10.1109/TNNLS.2016.2599820.
S. Otsuki dkk., “Layer-Wise Relevance Propagation with Conservation Property for ResNet,” Jul 2024, [Daring]. Tersedia pada: http://arxiv.org/abs/2407.09115
T. Bender dkk., “Analysis of a Deep Learning Model for 12-Lead ECG Classification Reveals Learned Features Similar to Diagnostic Criteria,” IEEE J Biomed Health Inform, vol. 28, no. 4, hlm. 1848–1859, Apr 2024, doi: 10.1109/JBHI.2023.3271858.
P. Wagner dkk., “PTB-XL, a large publicly available electrocardiography dataset,” Sci Data, vol. 7, no. 1, Des 2020, doi: 10.1038/s41597-020-0495-6.
A. Krizhevsky, I. Sutskever, dan G. E. Hinton, “ImageNet Classification with Deep Convolutional Neural Networks.” [Daring]. Tersedia pada: http://code.google.com/p/cuda-convnet/
Z. Wang, W. Yan, dan T. Oates, “Time series classification from scratch with deep neural networks: A strong baseline,” dalam 2017 International Joint Conference on Neural Networks (IJCNN), IEEE, Mei 2017, hlm. 1578–1585. doi: 10.1109/IJCNN.2017.7966039.
C. J. Anders, D. Neumann, W. Samek, K.-R. Müller, dan S. Lapuschkin, “Software for Dataset-wide XAI: From Local Explanations to Global Insights with Zennit, CoRelAy, and ViRelAy,” Feb 2023.
