Lazuardi, R. Arif Firdaus (2019) Deteksi Lokasi Tulang Manusia Pada Citra Ultrasound Berbasis Deep Learning Menggunakan CNN Arsitektur YoloV3. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
07111650052003-Master_Thesis.pdf Download (2MB) | Preview |
Abstract
Deteksi tulang manusia pada citra ultrasound memiliki tantangan yang cukup kompleks. Hal ini disebabkan, representasi pantulan gelombang suara yang dipancarkan oleh sensor B-scan USG tidak hanya menampilkan spesimen tulang saja, melainkan di dalamnya juga terdapat otot, jaringan lunak, dan bagian-bagian lain di bawah jaringan kulit. Oleh sebab itu dibutuhkan sebuah sistem yang dapat mengenali secara otomatis spesimen tulang pada citra USG. Penelitian ini mengimplementasikan pembelajaran sistem berbasis deep learning menggunakan metode convolutional neural network (CNN) dengan arsitektur YOLOv3. Hasil pelatihan sistem dengan threshold IoU 0.5 dapat mengenali objek tulang pada
〖mAP〗_50, 〖mAP〗_75, dan 〖mAP〗_(50:95) dengan masing-masing nilai sebesar 99.98, 97.68, dan 85.67. Dan untuk hasil pelatihan sistem dengan threshold IoU 0.75 dapat mengenali objek tulang pada 〖mAP〗_50, 〖mAP〗_75, dan 〖mAP〗_(50:95) dengan masing-masing nilai sebesar 99.96, 97.46, dan 86.35.
==================================================================================================================================
Detection of human bones in ultrasound images has quite complex challenges. This is due to the representation of reflections of sound waves emitted by the B-scan ultrasound sensors not only displaying bone specimens, but also muscle, soft tissue, and other parts under the skin tissue. Therefore we need a system that can automatically recognize bone specimens in ultrasound images. This study implements deep learning based systems using the convolutional neural network (CNN) method with the YOLOv3 architecture. The results of system training with IoU threshold 0.5 can recognize bone objects in〖mAP〗_50, 〖mAP〗_75, and 〖mAP〗_(50:95) with values of 99.98, 97.68, and 85.67 respectively. And for the results of system training with IoU 0.75 threshold can recognize bone objects in 〖mAP〗_50, 〖mAP〗_75, and 〖mAP〗_(50:95) with values of 99.96, 97.46, and 86.35 respectively.
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | RTE 006.32 Laz d-1 2019 |
| Uncontrolled Keywords: | USG tulang, CNN, YOLOv3 |
| Subjects: | Q Science > QA Mathematics > QA75 Electronic computers. Computer science. EDP Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis |
| Depositing User: | R. Arif Firdaus Lazuardi |
| Date Deposited: | 15 Jan 2025 07:20 |
| Last Modified: | 15 Jan 2025 07:20 |
| URI: | http://repository.its.ac.id/id/eprint/66415 |
Actions (login required)
 |
View Item |