Registrasi Citra Pembuluh Darah Otak terhadap Atlas Pembuluh Darah

Harvianti, Azizah Elok (2025) Registrasi Citra Pembuluh Darah Otak terhadap Atlas Pembuluh Darah. Project Report. [s.n.], [s.l.]. (Unpublished)

Text 5025221243-Project_Report.pdf - Accepted Version Restricted to Repository staff only Download (1MB) | Request a copy

Abstract

Perkembangan teknologi di bidang kesehatan telah memungkinkan pemanfaatan citra medis untuk mendukung proses diagnosis berbagai penyakit, termasuk aneurisma yang merupakan pelebaran abnormal pada pembuluh darah otak. Untuk mendeteksi kondisi tersebut secara akurat, diperlukan teknik pengolahan citra yang mampu memberikan representasi anatomi yang presisi, salah satunya melalui proses registrasi citra. Registrasi citra bertujuan menyelaraskan dua atau lebih citra dari waktu, sudut pandang, atau modalitas berbeda sehingga struktur anatomi dapat dibandingkan secara tepat. Dalam kerja praktik ini dilakukan implementasi dan evaluasi metode registrasi citra pembuluh darah otak menggunakan dataset Magnetic Resonance Imaging (MRI), dengan tujuan utama menyelaraskan citra pasien terhadap atlas pembuluh darah otak. Proses registrasi meliputi beberapa tahapan, mulai dari tahap preprocessing yang mencakup konversi DICOM ke NIfTI, reorientasi citra, dan penyesuaian orientasi, konfigurasi transformasi, hingga proses registrasi berbasis menggunakan beberapa optimizer, yaitu Powell, Regular Step Gradient Descent (RSGD), dan LBFGSB. Evaluasi kinerja dilakukan menggunakan metrik Mean Squared Error (MSE), Structural Structural Similarity Index Measure (SSIM), dan Mutual Information (MI) untuk mengukur tingkat kesesuaian antara citra hasil registrasi dan atlas. Selain itu, pengaruh penambahan noise juga diuji untuk menilai robustness sistem terhadap variasi kualitas citra. Hasil pengujian menunjukkan bahwa pada kondisi tanpa noise, optimizer Powell menghasilkan kinerja lebih baik dibandingkan RSGD dan LBFGSB, dengan nilai MSE rata-rata sebesar 2636,505, nilai SSIM sebesar 0,270, dan nilai MI sebesar 0,404. Hasil tersebut menunjukkan bahwa optimizer Powell menghasilkan nilai MSE terendah serta nilai SSIM dan MI tertinggi dibandingkan optimizer RSGD dan LBFGSB pada kondisi tanpa noise.
==================================================================================================================================
Advances in healthcare technology have enabled the use of medical imaging to support the diagnosis of various diseases, including aneurysms, which are abnormal dilations of cerebral blood vessels. Accurate detection of this condition requires image processing techniques capable of providing precise anatomical representations, one of which is achieved through image registration. Image registration aims to align two or more images acquired at different times, from different viewpoints, or using different modalities, so that anatomical structures can be accurately compared. In this internship project, the implementation and evaluation of cerebral blood vessel image registration methods were conducted using a Magnetic Resonance Imaging (MRI) dataset, with the primary objective of aligning patient images to a cerebral vascular atlas. The registration process consisted of several stages, starting with preprocessing steps that included DICOM-to-NIfTI conversion, image reorientation, and orientation adjustment, followed by transformation configuration and the registration process itself using several optimizers, such as Powell, Regular Step Gradient Descent (RSGD), and LBFGSB. The performance of each optimizer was evaluated using Mean Squared Error (MSE), Structural Similarity Index Measure (SSIM), and Mutual Information (MI) to measure the similarity between the registered images and the atlas. In addition, noise was added to the images to assess the robustness of the registration method under different image quality conditions. The experimental results showed that under noise-free conditions, the Powell optimizer achieved better performance compared to RSGD and LBFGSB, with an average MSE value of 2636,505, an SSIM value of 0,270, and an MI value of 0,404. These results indicate that the Powell optimizer produced the lowest MSE as well as the highest SSIM and MI values compared to the RSGD and LBFGSB optimizers under noise-free conditions.

Item Type:	Monograph (Project Report)
Uncontrolled Keywords:	Atlas Otak, Mutual Information, MRI, Noise, Powell, Registrasi Citra, Time-of-Flight (TOF)
Subjects:	Q Science Q Science > QH Biology Q Science > QM Human anatomy T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Azizah Elok Harvianti
Date Deposited:	07 Jan 2026 08:14
Last Modified:	07 Jan 2026 08:14
URI:	http://repository.its.ac.id/id/eprint/129344

Actions (login required)

View Item