Multilingual Medical Brain Image Captioning pada Brain Tumor Segmentation untuk Penutur Bahasa Indonesia

Fathir, Mirza Syahrizal (2025) Multilingual Medical Brain Image Captioning pada Brain Tumor Segmentation untuk Penutur Bahasa Indonesia. Project Report. [s.n.], [s.l.]. (Unpublished)

Text 5025231151-Project_Report.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Analisis citra medis merupakan aspek krusial dalam diagnosis tumor otak, tetapi interpretasi manual hasil pemindaian MRI memerlukan waktu dan ketelitian tinggi. Proyek ini mengembangkan Multilingual Medical Brain Image Captioning pada dataset Brain Tumor Segmentation (BraTS). Tujuan utama dari sistem ini adalah menghasilkan deskripsi teks otomatis mengenai kondisi tumor otak dalam Bahasa Indonesia untuk membantu tenaga medis dalam memahami karakteristik visual dari hasil pemindaian dalam Bahasa Indonesia. Arsitektur yang digunakan encoder-decoder terpisah. Bagian encoder berfungsi untuk mengekstraksi fitur spasial dari citra MRI tumor otak, sedangkan bagian decoder berbasis bahasa bertugas memprediksi caption yang relevan. Hasil dari proyek ini menunjukkan bahwa integrasi antara visi komputer dan pemrosesan bahasa alami dapat membantu menghasilkan caption berbahasa Indonesia dari citra MRI dengan keakuratan 6,8% pada metrik evaluasi BLEU-4.
====================================================================================================================================
Medical image analysis is a crucial aspect in brain tumor diagnosis, but manual interpretation of MRI scan results requires time and high accuracy. This project develops Multilingual Medical Brain Image Captioning on the Brain Tumor Segmentation (BraTS) dataset. The main objective of this system is to generate automatic text descriptions of brain tumor conditions in Indonesian to assist medical personnel in understanding the visual characteristics of scan results in Indonesian. The architecture used is a separate encoder-decoder. The encoder part functions to extract spatial features from brain tumor MRI images, while the language-based decoder part is tasked with predicting relevant captions. The results of this project show that the integration between computer vision and natural language processing can help generate Indonesian captions from MRI images with an accuracy of 6.8% on the BLEU-4 evaluation metric.

Item Type:	Monograph (Project Report)
Uncontrolled Keywords:	Bahasa Indonesia, Brain, Caption, Encoder- Decoder, Otak, Takarir, Indonesian.
Subjects:	R Medicine > R Medicine (General) > R858 Deep Learning R Medicine > RC Internal medicine > RC78.7.N83 Magnetic resonance imaging.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Mirza Syahrizal Fathir
Date Deposited:	12 Jan 2026 02:45
Last Modified:	12 Jan 2026 02:45
URI:	http://repository.its.ac.id/id/eprint/129440

Actions (login required)

View Item