Augmentasi Citra Ultrasound Pembuluh Darah menggunakan Generative Adversarial Networks

Sulistyo, Pralambang Manggala (2023) Augmentasi Citra Ultrasound Pembuluh Darah menggunakan Generative Adversarial Networks. Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 07211840000062-Undergraduate_Thesis.pdf] Text
07211840000062-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 April 2025.

Download (10MB) | Request a copy

Abstract

Deep learning adalah salah satu jenis dari machine learning yang bertugas untuk melatih komputer agar bisa melakukan pekerjaan seperti manusia, seperti mendeteksi ucapan, mengidentifikasi gambar, dan membuat prediksi. Namun dalam model Deep Learning memiliki kelemahan yaitu proses pelatihan model cukup lama. Masalah yang paling sering muncul adalah kurangnya jumlah data pelatihan yang cukup atau keseimbangan kelas data yang tidak merata dalam dataset. Sebagaimana yang telah diketahui bahwa lebih banyak data yang diakses akan lebih efektif hasil akurasinya. Saat ini untuk penelitian medis menggunakan Deep Learning terkendala pada kurangnya jumlah data pelatihan yang cukup atau keseimbangan kelas data yang tidak merata dalam data, sebagai contohnya adalah Citra Ultrasound Pembuluh Darah. Oleh karena itu diperlukannya Data Augmentation pada Citra Medis tersebut. Untuk melakukan Data Augmentation pada citra data medis, dibutuhkan suatu sistem untuk meng augmentasi beberapa sample citra real secara otomatis. Digunakan metode Supervised Data Augmentation yaitu Generative Adversarial Networks. GAN dilatih untuk mampu membangkitkan suatu gambar baru berdasarkan kumpulan gambar yang telah ia lihat sebelumnya selama proses pelatihan. Hasil yang diharapkan dari tugas akhir ini adalah berhasil menerapkan Data Augmentation dengan Generative Adversarial Networks kepada Citra Ultrasonografi Pembuluh Darah.
======================================================================================================================================
Deep learning is a type of machine learning that is used to train computers to do jobs like humans, such as detecting speech, identifying images, and making predictions. However, the Deep Learning model has a weakness, namely the model training process is quite long. The most common problems are the lack of sufficient training amounts or the uneven balance of data classes in the data set. As it is known that the more data that is accessed, the more effective the accuracy results will be. Currently, medical research using Deep Learning is constrained by the lack of sufficient training or the uneven balance of data classes in the data, for example, Ultrasound Image of Blood Vessels. Therefore, data augmentation is needed on the medical image. To perform Data Augmentation on medical data images, a system is needed to augment several real image samples automatically. The method used is Supervised Data Augmentation, namely Generative Adversarial Networks. The GAN knows to generate a new image the set of images it has seen previously during the process. The expected result of this final project is to successfully apply Data Augmentation with Generative Adversarial Networks to Ultrasonographic Image of Blood Vessels.

Item Type: Thesis (Other)
Uncontrolled Keywords: Deep Learning, Generative Adversarial Networks, Medical Image, Data Augmentation, Ultrasonographic, Deep Learning, Generative Adversarial Networks, Citra Medis, Data Augmentation.
Subjects: R Medicine > R Medicine (General) > R858 Deep Learning
T Technology > T Technology (General) > T57.5 Data Processing
T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User: Pralambang Manggala Sulistyo
Date Deposited: 17 Feb 2023 06:29
Last Modified: 17 Feb 2023 06:29
URI: http://repository.its.ac.id/id/eprint/97525

Actions (login required)

View Item View Item