Clarasanty, Anindya (2025) Segmentasi Citra Ultrasonografi Arteri Karotis Berbasis Deep Learning Untuk Analisis Karakter Elastisitas Pembuluh Darah. Other thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
5023211005-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (6MB) | Request a copy |
Abstract
Cardiovascular Disease (CVD) merupakan penyebab utama kematian, dengan jumlah kematian mencapai 17,9 juta jiwa setiap tahunnya. Salah satu faktor risiko yang mendasari perkembangan CVD adalah vascular aging, yaitu penurunan elastisitas pembuluh darah seiring bertambahnya usia. Identifikasi dini kondisi vaskular sangat penting untuk mencegah komplikasi lebih lanjut. Pencitraan ultrasonografi yang dimodifikasi dengan sensor tekanan menawarkan solusi yang lebih terjangkau, aman, dan mudah diakses untuk memantau kondisi vaskular melalui analisis elastisitas arteri. Segmentasi citra ultrasonografi arteri karotis yang akurat diperlukan untuk mengekstraksi fitur terkait perubahan diameter arteri. Dalam penelitian ini, metode segmentasi menggunakan model deep learning UNet-ResNet34 diusulkan untuk meningkatkan akurasi dan efisiensi pemrosesan. Model ini mencapai nilai IoU sebesar 0,9545 dengan F1-score sebesar 0,9988 pada ambang IoU 0,5 selama proses pelatihan, serta IoU sebesar 0,6101 dan F1-score sebesar 0,7907 saat diuji pada data subjek. Evaluasi pada data uji juga dilakukan dengan membandingkan diameter tinggi arteri yang diprediksi dengan ground truth yang dibuat secara manual, menghasilkan mean absolute error (MAE) sebesar 1,46 mm dan mean relative error (MRE) sebesar 24,37%. Subjek berusia 18–25 tahun memiliki nilai Young’s Elastic Modulus (YEM) berkisar antara 30–50 kPa, sementara subjek berusia 29–54 tahun menunjukkan peningkatan nilai YEM yang melebihi 100 kPa. Selain nilai YEM, perubahan diameter tinggi arteri juga terlihat menurun pada individu yang lebih tua. Subjek berusia 18–25 tahun menunjukkan perubahan tinggi lebih dari 0,2 mm, sedangkan mereka yang berusia 33 tahun ke atas hanya mengalami perubahan sekitar ≤ 0,1 mm. Perubahan tinggi arteri pada subjek muda sekitar dua kali lebih besar, menunjukkan bahwa arteri karotis pada individu yang lebih muda lebih elastis dan mampu mengalami distensi yang lebih besar terhadap tekanan mekanik dibandingkan dengan individu yang lebih tua.
======================================================================
Cardiovascular Disease (CVD) is the leading cause of death, accounting for 17.9 million deaths each year. One of the underlying risk factors for the development of CVD is vascular aging, which refers to the progressive decline in blood vessel elasticity with increasing age. Early identification of vascular health conditions is essential to prevent further complications. Ultrasound imaging modified with a pressure sensor offers a more accessible, safe, and affordable solution to monitor vascular conditions by analyzing arterial elasticity. Accurate segmentation of carotid artery ultrasound images is needed to extract features related to arterial diameter changes. In this study, a segmentation method using the deep learning model UNet-ResNet34 is proposed to improve accuracy and processing efficiency. The model achieved an IoU of 0.9545 with an F1-score of 0.9988 at an IoU threshold of 0.5 during training, and an IoU of 0.6101 with an F1-score of 0.7907 when tested on subject data. Evaluation on test data was also conducted by comparing the predicted arterial height diameter with the manually created ground truth, resulting in a mean absolute error (MAE) of 1.46 mm and a mean relative error (MRE) of 24.37%. Individuals aged 18–25 years had Young’s Elastic Modulus (YEM) values ranging from 30–50 kPa, while those aged 29–54 years showed increased values exceeding 100 kPa. In addition to YEM, the change in arterial height diameter was observed to decrease in older individuals. Subjects aged 18–25 years exhibited height changes greater than 0.2 mm, whereas those aged 33 and above experienced changes of approximately ≤ 0.1 mm. The height change in younger subjects was roughly twice as large, indicating that carotid arteries in younger individuals are more elastic of greater distension in response to mechanical pressure compared to older individuals.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Arteri, CVD, Deep Learning, Segmentasi, Ultrasonografi Artery, CVD, Deep Learning, Segmentation, Ultrasound |
Subjects: | R Medicine > R Medicine (General) > R856.2 Medical instruments and apparatus. R Medicine > R Medicine (General) > R858 Deep Learning R Medicine > RC Internal medicine > RC691 Blood-vessels--Diseases. R Medicine > RC Internal medicine > RC78.7.U4 Ultrasonic imaging. T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.62 Decision support systems T Technology > T Technology (General) > T59.7 Human-machine systems. |
Divisions: | Faculty of Electrical Technology > Biomedical Engineering > 11410-(S1) Undergraduate Thesis |
Depositing User: | Anindya Clarasanty |
Date Deposited: | 04 Aug 2025 04:43 |
Last Modified: | 04 Aug 2025 04:43 |
URI: | http://repository.its.ac.id/id/eprint/126064 |
Actions (login required)
![]() |
View Item |