Bontong, Wean Harmiwan (2026) Sistem Skrining Kondisi Pendengaran Menggunakan Transient-Evoked Otoacoustic Emissions Berbasis Rule-Based. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5023221041-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5023221041-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy |
Abstract
Gangguan pendengaran merupakan salah satu disabilitas sensorik yang banyak ditemukan di masyarakat dan dapat menimbulkan dampak yang signifikan terhadap kualitas hidup. Metode skrining pendengaran yang umum digunakan, seperti audiometri nada murni, masih memerlukan respons subjek sehingga bersifat subjektif, sedangkan metode objektif seperti Auditory Brainstem Response (ABR) relatif mahal dan memerlukan waktu pemeriksaan yang lebih lama. Di sisi lain, Transient-Evoked Otoacoustic Emissions (TEOAE) menawarkan metode evaluasi fungsi koklea yang objektif, cepat, dan non-invasif, namun perangkat komersial yang tersedia masih memiliki biaya yang tinggi. Oleh karena itu, penelitian ini bertujuan mengembangkan sistem skrining kondisi pendengaran berbasis TEOAE yang objektif, berbiaya rendah, dan mampu melakukan klasifikasi otomatis menggunakan pendekatan rule-based. Sistem dikembangkan menggunakan probe rakitan sendiri (DIY) berbasis speaker dan mikrofon sensitivitas tinggi yang terintegrasi dengan ESP32 dan Raspberry Pi Zero. Stimulus yang digunakan berupa Derived Non-Linear Response (DNLR) untuk membangkitkan respons TEOAE. Sinyal yang diperoleh diproses melalui tahap averaging dan pemfilteran domain waktu-frekuensi untuk meningkatkan kualitas sinyal. Validitas perekaman dievaluasi menggunakan Cross Spectral Coherence (CSC), sedangkan klasifikasi kondisi pendengaran dilakukan menggunakan kombinasi parameter Whole Waveform Reproducibility (WWR), Signal-to-Noise Ratio (SNR), dan Two-Dimensional Radius (RAD2D). Hasil pengujian menunjukkan bahwa sistem mampu melakukan skrining kondisi pendengaran dengan sensitivitas sebesar 95,45%, spesifisitas sebesar 100%, dan akurasi sebesar 97,73%. Hasil tersebut menunjukkan bahwa kombinasi parameter WWR, SNR, dan RAD2D mampu meningkatkan kemampuan sistem dalam membedakan pendengaran normal dan gangguan pendengaran. Dengan demikian, sistem yang dikembangkan berpotensi menjadi alternatif alat skrining pendengaran yang objektif dengan harga yang terjangkau untuk mendukung deteksi dini gangguan pendengaran.
=========================================================================================================================================
Hearing loss is one of the most common sensory disabilities in the general population and can have a significant impact on quality of life. Commonly used hearing screening methods, such as pure-tone audiometry, still require a response from the subject and are therefore subjective, while objective methods such as Auditory Brainstem Response (ABR) are relatively expensive and require a longer examination time. On the other hand, Transient-Evoked Otoacoustic Emissions (TEOAE) offer an objective, rapid, and non-invasive method for evaluating cochlear function; however, commercially available devices remain expensive. Therefore, this study aims to develop a TEOAE-based hearing screening system that is objective, low-cost, and capable of automatic classification using a rule-based approach. The system was developed using a do-it-yourself (DIY) probe based on a speaker and a high-sensitivity microphone integrated with an ESP32 and a Raspberry Pi Zero. The stimulus used was a Derived Non-Linear Response (DNLR) to elicit TEOAE responses. The acquired signals were processed through averaging and time-frequency domain filtering to improve signal quality. The validity of the recordings was evaluated using Cross Spectral Coherence (CSC), while hearing condition classification was performed using a combination of the Whole Waveform Reproducibility (WWR), Signal-to-Noise Ratio (SNR), and Two-Dimensional Radius (RAD2D) parameters. Test results showed that the system was capable of screening for hearing conditions with a sensitivity of 95.45%, a specificity of 100%, and an accuracy of 97.73%. These results indicate that the combination of WWR, SNR, and RAD2D parameters enhances the system’s ability to distinguish between normal hearing and hearing loss. Thus, the developed system has the potential to serve as an affordable, objective alternative for hearing screening to support the early detection of hearing loss.
Actions (login required)
 |
View Item |