Klasifikasi Multi Kelas Untuk Mendeteksi Sinyal Detak Jantung Janin Pada CTG Menggunakan Thresholding Feature Importance dan ResNet Multimodal
DOI:
https://doi.org/10.57203/session.v4i1.2025.44-54Kata Kunci:
cardiotocography, fetal heart rate, multi-class classification, thresholding feature importance, multimodal ResNetAbstrak
Metode konvensional untuk memantau kondisi janin melalui rekaman denyut nadi janin (FHR) dan kontraksi uterus (UC), cardiotocography (CTG), dilakukan secara manual. Interpretasi manual CTG sering menyebabkan variasi antar pengamat dan meningkatkan kemungkinan kesalahan dalam mengidentifikasi hipoksia janin. Untuk memecahkan masalah tersebut, penelitian ini menyarankan teknik klasifikasi multi-kelas (Normal, Suspect, dan Hipoksia) yang menggunakan kombinasi ResNet multimodal dan Thresholding Importance Feature. Sebelum proses sinyal FHR, sampel, normalisasi, dan segmentasi berbasis time-window segmentation dilakukan. Selanjutnya, fitur klinis dan statistik seperti STV, LTV, akselerasi, deselerasi, pH, dan skor Apgar diekstrak. Selanjutnya, metode yang menggabungkan nilai gain, permutation, dan pilihan stabilitas digunakan untuk memilih stabilitas dengan nilai fitur terendah. Tiga cabang terdiri dari arsitektur multimodal yang dikembangkan: 1D-ResNet untuk sinyal FHR mentah; 2D-ResNet untuk representasi citra (time-frequency/spectrogram); dan MLP untuk data tabular hasil ambang. Ketiga cabang difusi melalui lapisan yang benar-benar terhubung untuk menghasilkan prediksi akhir multi-kelas. Hasil uji menunjukkan bahwa model memiliki akurasi 84,6%, skor F1 83, dan skor AUROC di atas 0,9. Kelas Suspect menunjukkan sensitivitas yang lebih tinggi, dan Kelas Normal menunjukkan prestasi terbaik. Studi ini menunjukkan bahwa fusi multimodal dengan thresholding fitur dapat meningkatkan keandalan sistem klasifikasi CTG multi-kelas dan memberikan kontribusi berupa pipeline hibrid yang efisien dan relevan klinis. Sebaliknya, kurva ROC dan analisis confusion matrix menunjukkan kemampuan diskriminatif yang baik. Tingkat risiko evaluasi mencapai 40%, yang sesuai dengan rekomendasi klinis
Referensi
[1] A. Mohan and V. Uma, “Fetal Hypoxia Detection Using CTG Signals and CNN Models,” International Research Journal of Advanced Science Hub (IRJASH), vol. 5, no. 5S, pp. 434–441, 2023.
[2] Y. D. Daydulo, B. L. Thamineni, H. K. Dasari, and G. T. Aboye, “Deep Learning Based Fetal Distress Detection from Time–Frequency Representation of Cardiotocogram Signal Using Morse Wavelet,” Authorea Preprint, Apr. 2022, doi: 10.22541/au.164973963.38287380/v1.
[3] K.-S. Lee et al., “Real-time Classification of Fetal Status Based on Deep Learning and Cardiotocography Data,” Journal of Medical Systems, vol. 47, no. 1, pp. 1–12, 2023.
[4] R. A. Alqahtani, G. N. Alotibi, and T. N. Alotaiby, “Fetal Hypoxia Classification from Cardiotocography Signals Using Instantaneous Frequency and Common Spatial Pattern,” Electronics, vol. 14, no. 5, p. 950, 2025.
[5] Nicole Chiou1, et al., Development and evaluation of deep learning models for cardiotocography interpretation, npj | women's health, https://doi.org/10.1038/s44294-025-00068-w
[6] D. Monica and R. Sulastri, “Klasifikasi Hasil Cardiotocography (CTG) Ibu Hamil untuk Memprediksi Kesehatan Janin,” Jurnal Teknologi dan Sistem Informasi, vol. 9, no. 2, pp. 45–52, 2023.
[7] F. Fahmi and A. Yulian, “REKOMENDASI KESEHATAN JANIN DENGAN PENERAPAN ALGORITMA C5.0 MENGGUNAKAN CLASSIFYING CARDIOTOCOGRAPHY DATASET,” Jurnal SIMANTEC, vol. 9, no. 2, P-ISSN : 2088-2130E-ISSN : 2502-4884, 2 Juni 2021
[8] A. Pradipta and A. P. Ayu, “Klasifikasi Fetal Cardiotocography Menggunakan Pendekatan Boosting Classifier,” JURNAL SISTEM DAN INFORMATIKA (JSI), vol. 12, no. 1, pp. 65–74, 2023. p-ISSN: 1858-473X, e-ISSN: 2460-3732
[9] A. Subairi, A. Rachman, and R. Santoso, “Detecting Sleep Apnea Using ECG Signal with Decision Tree,” Jurnal EECCIS Vol. 16, No. 3, Desember 2022, pp96-100, https://jurnaleeccis.ub.ac.id/e-ISSN(Online): 2460-8122
[10] A Venkata Sriram, et al, “Cardiotocogram Data classification using Machine Learning Algorithms along with Swarm-Based Metaheuristic optimizations for Autonomous Fetal distress detection, J. Electrical Systems 20-10s (2024): 6098-6113, January 2024
[11] Akshay Harikumar, et al, “Explainable AI in Deep Learning Based Classification of Fetal Ultrasound Image Planes,” ICIDCA 2024, Communication Technologies and Application 10.1016/j.procs.2024.03.291.
[12] Yefei Zhang, et al, “Multimodal learning for fetal distress diagnosis using a multimodal medical information fusion framework, vol. 145, p. 105409, 2022.
[13] S. Ali and F. Ahmed, “Feature Selection and Deep Residual Networks for CTG Signal Classification,” Front Physiol, 2022 Nov 7;13:1021400. doi: 10.3389/fphys.2022.1021400.
[14] Rawad A. Alqahtani, et al, “Fetal Hypoxia Classification from Cardiotocography Signals Using Instantaneous Frequency and Common Spatial Pattern, Electronics 2025, 14(5), 950; https://doi.org/10.3390/electronics14050950.
[15] Yared Daniel, et al, “Deep Learning Based Fetal Distress Detection from Time Frequency Representation of Cardiotocogram Signal Using Morse Wavelet,” April 2022DOI: 10.22541/au.164973963.38287380/v1
[16] D. C. Junyuan Feng et al., “A Hybrid Stacked Ensemble and Kernel SHAP-Based Model for Intelligent Cardiotocography Classification and Interpretability,” BMC Medical Informatics and Decision Making volume 23, Article number: 273 (2023)
[17] Yue Fei, et al, “Intelligent classification of antenatal cardiotocography signals via multimodal bidirectional gated recurrent units,” 2022, Biomedical Signal Processing and Control, p. 104008, DOI, 10.1016/j.bspc.2022.104008
[18] Ramdas Kapila, et al, “Optimizing fetal health prediction: Ensemble modeling with fusion of feature selection and extraction techniques for cardiotocography data,” Computational Biology and Chemistry Volume 107, December 2023, 107973 3.
[19] Farah Francis, et al, “Machine learning on cardiotocography data to classify fetal outcomes: A scoping review,” Comput Biol Med 2024 Apr:172:108220. doi: 10.1016/j.compbiomed.2024.108220. Epub 2024 Mar 7.
[20] Tasnim Bill Zannah, et al, “Comparative performance analysis of ensemble learning methods for fetal health classification,” Informatics in Medicine Unlocked Volume 56, 2025, 101656
[21] Subono1, Alfin Hidayat2, wardhany, Vivien Arief3, et al, “Doppler Ultrasound Sense Performance for Cardiotography (CTG) Fetal Baby Heartbeat Detector Using Arduino Mega Pro Mini,” Proceeding - 6th International Conference on Information Technology, Information Systems and Electrical Engineering: Applying Data Sciences and Artificial Intelligence Technologies for Environmental Sustainability, ICITISEE 2022 Conference Paper 2022 DOI: 10.1109/ICITISEE57756.2022.10057788
[22] Subono Subono1,Sholeh Hadi Pramono, Erny Yudaningtyas, Muhammad Aziz Muslim “Fetal Heart Rate Data Augmentation Using DCGAN for Multiclass Classification in Normal, Suspected, and Pathological Conditions,” Conference: 2025 International Conference on Computer Sciences, Engineering, and Technology Innovation (ICoCSETI), January 2025 DOI: 10.1109/ICoCSETI63724.2025.11019225
Unduhan
Diterbitkan
Terbitan
Bagian
Lisensi
Hak Cipta (c) 2025 Subono Subono, Mohamad Dimyati Ayatullah, Sholeh Hadi Pramono, Erni Yudaningtyas, M. Aziz Muslim, Mahdin Rohmatillah, Cries Avian
Artikel ini berlisensi Creative Commons Attribution 4.0 International License.
Copyright :
Authors who publish their manuscripts in this journal agree to the following conditions.
- The copyright on each article belongs to the author(s).
- The author acknowledges that Jurnal SESSION: Software Development, Digital Business Intelligence, and Computer Engineering has the right to be the first to publish with a Creative Commons Attribution 4.0 International license (Attribution 4.0 International (CC BY 4.0).
- Authors can submit articles separately, arrange for the nonexclusive distribution of manuscripts that have been published in this journal into other versions (e.g., sent to the author's institutional repository, publication into books, etc.), by acknowledging that the manuscript has been published for the first time in the Jurnal SESSION: Software Development, Digital Business Intelligence, and Computer Engineering.
License:
Jurnal SESSION: Software Development, Digital Business Intelligence, and Computer Engineering is licensed under a Creative Commons Attribution 4.0 International License.
This license permits anyone to copy and redistribute this material in any form or format, compose, modify, and make derivatives of this material for any purpose, including commercial purposes, as long as the author credits the original work.
