Segmentasi Arteri Karotis Pada Citra Ultrasound Berbasis Deep Learning Untuk Estimasi Elastisitas Pembuluh Darah

Mujahid, Muhammad Ridho (2025) Segmentasi Arteri Karotis Pada Citra Ultrasound Berbasis Deep Learning Untuk Estimasi Elastisitas Pembuluh Darah. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5024211023-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (30MB) | Request a copy

Abstract

Penelitian ini bertujuan untuk mengembangkan sistem berbasis deep learning untuk melakukan segmentasi lumen arteri karotis pada citra ultrasound serta mengestimasi elastisitas pembuluh darah secara non-invasif. Model segmentasi dibangun menggunakan arsitektur UNet dengan kombinasi preprocessing speckle filtering dan augmentasi geometris, dilatih pada dataset publik Common Carotid Artery Ultrasound Images. Selain itu, sistem akuisisi data nyata dirancang dengan menambahkan sensor tekanan berbasis load cell pada probe ultrasound, yang memungkinkan sinkronisasi antara video citra B-mode dan data tekanan. Hasil segmentasi lumen digunakan untuk menghitung diameter arteri setiap frame, yang kemudian dianalisis terhadap tekanan untuk menghitung parameter elastisitas seperti distensibility coefficient, strain, Peterson’s elastic modulus (PEM), dan stiffness index β. Model terbaik menunjukkan nilai Dice sebesar 0,931 dan IoU sebesar 0,871, serta hasil estimasi elastisitas yang berada dalam rentang fisiologis menurut literatur. Sistem ini menunjukkan potensi untuk digunakan sebagai alat bantu analisis kesehatan pembuluh darah secara efisien dan non invasif.
======================================================================================================================================
This study aims to develop a deep learning-based system for segmenting the carotid artery lumen in ultrasound images and estimating vascular elasticity non-invasively. The segmentation model was built using a U-Net architecture combined with speckle filtering and geometric augmentation, and was trained on the publicly available Common Carotid Artery Ultrasound Images dataset. Additionally, a real-time acquisition system was designed by integrating a load cell pressure sensor into the ultrasound probe, allowing synchronization between B-mode video frames and pressure data. The segmented lumen was used to calculate artery diameter on a perframe basis, which was then analyzed against contact pressure to estimate elasticity parameters such as distensibility coefficient, strain, Peterson’s elastic modulus (PEM), and stiffness index β. The best-performing model achieved a Dice score of 0.931 and an IoU of 0.871, with elasticity estimates falling within physiologically accepted ranges according to existing literature. This system demonstrates potential as an efficient and non invasive tool for vascular health analysis.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Segmentasi, Deep Learning, Citra Ultrasound, Arteri Karotis, Elastisitas Pembuluh Darah, Segmentation, Deep Learning, Ultrasound Imaging, Carotid Artery, Vascular Elasticity.
Subjects:	R Medicine > R Medicine (General) > R858 Deep Learning T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	Muhammad Ridho Mujahid
Date Deposited:	30 Jul 2025 07:10
Last Modified:	30 Jul 2025 07:10
URI:	http://repository.its.ac.id/id/eprint/123245

Actions (login required)

View Item