Shodiq, Moh. Nur (2026) Estimasi Volume Thrombus Berbasis Citra Ultrasound Ganda Berorientasi Menggunakan Deep Learning. Doctoral thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 7022211012-Doctoral.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
7022211012-Doctoral.pdf - Accepted Version Restricted to Repository staff only Download (15MB) | Request a copy |
Abstract
Deep Vein Thrombosis merupakan kondisi terbentuknya gumpalan darah abnormal di dalam pembuluh vena bagian dalam, umumnya pada tungkai bawah, yang dapat menimbulkan risiko kesehatan serius. Penanganan konvensional terhadap DVT umumnya dilakukan melalui prosedur penyedotan gumpalan darah yang dipandu dengan angiografi sinar-X, namun metode ini memiliki kelemahan berupa paparan radiasi terhadap pasien dan tenaga medis. Penelitian ini bertujuan untuk meningkatkan akurasi diagnosis dan efektivitas penanganan DVT melalui pengembangan metode rekonstruksi 3D menggunakan citra ultrasonografi, interpolasi linear 3D, serta pendekatan multidenoising filter untuk meningkatkan kualitas segmentasi citra. Metodologi penelitian mencakup akuisisi data ultrasonografi menggunakan pemindai B-mode yang dikombinasikan dengan sistem pelacakan optik, diikuti dengan proses rekonstruksi volume 3D melalui teknik bin-filling dan hole-filling. Selanjutnya, teknik deep learning diterapkan untuk melakukan segmentasi area trombus pada citra ultrasonografi, serta memperkirakan volume trombus secara kuantitatif. Eksperimen dilakukan dalam dua skenario utama, yaitu (1) menggunakan dataset citra ultrasonografi 2D dari pasien DVT, dan (2) penentuan area trombus menggunakan dataset buatan yang terdiri atas fat-injected balloon phantom. Temuan ini menunjukkan bahwa metode yang dikembangkan berpotensi menjadi alternatif yang aman dan efektif untuk merekonstruksi volume trombus serta mengidentifikasi area trombus pada citra ultrasonografi, sekaligus menawarkan solusi tanpa paparan radiasi dibandingkan metode berbasis sinar-X konvensional.
============================================================================================================================
Deep Vein Thrombosis refers to the formation of abnormal blood clots within the deep venous system, commonly in the lower limbs, which can pose serious health risks. Conventional treatment of DVT generally involves thrombus suction guided by X-ray angiography; however, this method exposes both patients and medical personnel to radiation. This study aims to improve the accuracy of diagnosis and the effectiveness of DVT management by developing a three-dimensional (3D) reconstruction method using ultrasound imaging, 3D linear interpolation, and a multidenoising filter approach to enhance image segmentation quality. The research methodology includes ultrasound data acquisition using a B-mode scanner integrated with an optical tracking system, followed by 3D volume reconstruction through bin-filling and hole-filling techniques. Subsequently, a deep learning approach is employed to segment the thrombus area in ultrasound images and to quantitatively estimate the thrombus volume. Experiments were conducted in two main scenarios: (1) using 2D ultrasound image datasets obtained from DVT patients, and (2) thrombus area determination using artificial datasets composed of fat-injected balloon phantoms. The findings demonstrate that the proposed method has strong potential as a safe and effective alternative for reconstructing thrombus volume and identifying thrombus regions in ultrasound images, while eliminating radiation exposure associated with conventional X-ray-based methods.
| Item Type: | Thesis (Doctoral) |
|---|---|
| Uncontrolled Keywords: | Thrombosis Vena Dalam (TVD), rekonstruksi 3D, pen- citraan ultrasonografi, penyaring multi-denoising, pembelajaran mendalam, segmentasi trombus, pelacakan optik, Deep Vein Thrombosis, 3D reconstruction, ultrasound image, multi-denoising filter, deep learning, thrombus segmentation, optical tracking |
| Subjects: | 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) > Electrical Engineering > 20001-(S3) PhD Thesis |
| Depositing User: | Shodiq Moh. Nur |
| Date Deposited: | 21 Jan 2026 08:02 |
| Last Modified: | 21 Jan 2026 08:02 |
| URI: | http://repository.its.ac.id/id/eprint/129807 |
Actions (login required)
 |
View Item |