Sistem Deteksi Atrial Fibrillation (AF) Menggunakan Radar Continuous Wave (CW) 24 GHz

Hakim, Muhammad Rusydi Al (2026) Sistem Deteksi Atrial Fibrillation (AF) Menggunakan Radar Continuous Wave (CW) 24 GHz. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5023221002-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Atrial Fibrillation (AF) merupakan aritmia yang paling sering dijumpai secara klinis dan berkontribusi signifikan terhadap peningkatan risiko stroke serta gagal jantung, sehingga deteksi dini menjadi aspek yang sangat penting. Metode diagnosis konvensional berbasis electrocardiogram (ECG) memerlukan kontak langsung dengan tubuh, sehingga kurang ideal untuk pemantauan jangka panjang maupun pada individu dengan kondisi kulit tertentu. Penelitian ini mengembangkan sistem deteksi AF non-kontak menggunakan radar Continuous Wave (CW) 24 GHz dengan mengoptimalkan pengolahan sinyal radar untuk memperoleh parameter Heart Rate Variability (HRV) yang optimal sebagai dasar klasifikasi kondisi AF dan non-AF. Sistem dilengkapi mikrokontroler STM32F303K8T6 untuk akuisisi simultan sinyal radar dan ECG sebagai acuan (gold standard), serta rangkaian pengondisi sinyal dengan rentang passband 0,0891–3,968 Hz yang sesuai dengan spektrum fisiologis sinyal jantung dan mampu melakukan akuisisi tanpa lag (0 ms). Sinyal radar didemodulasi dengan menggunakan perbandingan metode arctangent dan metode Modified Differentiate and Cross-Multiply (MDACM), kemudian interval antar denyut (BBI) diekstraksi untuk memperoleh fitur HRV pada domain waktu, domain frekuensi, dan domain non-linier. Validasi HRV radar terhadap HRV ECG menunjukkan nilai RMSE sebesar 56 ms (MAE 43 ms) pada kondisi normal dan 100 ms (MAE 79 ms) pada kondisi AF. Didapatkan hasil klasifikasi menggunakan Logistic Regression dengan hasil akurasi 89,92 %, sensitivity 89,90 %, specifity 90,12 %, precision 90,14 % dan metrik F1 serta AUC masing-masing 90,07 % dan 94,78 %. Analisis HRV menunjukkan peningkatan pada parameter SDNN dan RMSSD, pelebaran elips pada plot Poincaré, serta peningkatan daya LF dan HF pada kondisi AF yang mencerminkan tingginya variabilitas interval RR. Hasil penelitian ini menunjukkan bahwa radar CW 24 GHz dapat mendukung implementasi pemantauan kesehatan jangka panjang yang lebih nyaman dan efisien.
======================================================================================================================================
Atrial Fibrillation (AF) is the most common cardiac arrhythmia encountered in clinical practice and is strongly associated with an increased risk of stroke and heart failure, making early detection essential. Conventional diagnostic methods based on electrocardiography (ECG) require direct skin contact, limiting their suitability for long-term monitoring and for individuals with certain skin conditions. This study presents a non-contact AF detection system using a 24 GHz Continuous Wave (CW) radar, with optimized radar signal processing to extract reliable Heart Rate Variability (HRV) parameters for distinguishing between AF and non-AF conditions. The developed system integrates an STM32F303K8T6 microcontroller for the simultaneous acquisition of radar and ECG signals, where the ECG serves as the gold standard. A signal conditioning circuit with a passband of 0.0891–3.968 Hz, matching the physiological frequency range of cardiac signals, was designed to enable signal acquisition without measurable latency (0 ms). Radar signals were demodulated using both the arctangent and Modified Differentiate and Cross-Multiply (MDACM) methods, followed by beat-to-beat interval (BBI) extraction to derive HRV features in the time, frequency, and nonlinear domains. Validation against ECG-derived HRV yielded an RMSE of 56 ms (MAE 43 ms) under normal rhythm and 100 ms (MAE 79 ms) during AF. The Logistic Regression classifier achieved an accuracy of 89,92%, sensitivity of 89,90%, specificity of 90,12%, precision of 90,14%, F1-score of 90.07%, and an AUC of 94,78%. HRV analysis further revealed increased SDNN and RMSSD values, a wider Poincaré plot ellipse, and elevated LF and HF spectral power during AF, reflecting the greater variability of RR intervals associated with the arrhythmic condition. These findings demonstrate the potential of a 24 GHz CW radar as a comfortable and efficient non-contact solution for long-term cardiovascular monitoring and AF detection.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Atrial Fibrillation, Radar 24 GHz, Machine Learning, Heart Rate Variability; Atrial Fibrillation, 24 GHz Radar, Machine Learning, Heart Rate Variability
Subjects:	T Technology > T Technology (General) T Technology > T Technology (General) > T57.5 Data Processing T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5102.9 Signal processing.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Biomedical Engineering > 11410-(S1) Undergraduate Thesis
Depositing User:	Muhammad Rusydi Al Hakim
Date Deposited:	01 Aug 2026 06:17
Last Modified:	01 Aug 2026 06:17
URI:	http://repository.its.ac.id/id/eprint/141403

Actions (login required)

View Item