Skripsi

IMAGE CAPTIONING BERBASIS DEEP LEARNING PADA CITRA REUMATOLOGI

---

The advancement of medical imaging technology has produced increasingly complex data that require fast and accurate analysis. However, the interpretation of medical images still requires considerable time, relies heavily on medical professionals, and is prone to interpretation errors. Therefore, this study aims to develop an Image Captioning system capable of automatically generating descriptions for medical images. The dataset consists of 121 rheumatology ultrasound images of the knee joint categorized into three classes. This study proposes an Image Captioning model based on an encoder–decoder architecture using three encoders, namely DenseNet121, ResNet101, and ViTBase, and two decoders, namely LSTM and Transformer Block. The models were trained for 50 epochs with a batch size of 12 and evaluated using the BLEU, METEOR, and ROUGE-L metrics. The experimental results show that the Image Captioning model with the DenseNet121–LSTM architecture achieved the best performance in generating descriptions for rheumatology images, obtaining the highest scores of 0.93 for BLEU-4, 0.972 for METEOR, and 0.973 for ROUGE-L. Keywords: Deep Learning, Image Captioning, Rheumatology, DenseNet, ResNet, Transformer, LSTM.

---

Availability

Detail Information

Series Title

-

Call Number

T2009632026

Publisher

Indralaya : Prodi Sistem Komputer, Fakultas Ilmu Komputer Universitas Sriwijaya., 2026

Collation

xvii, 106 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

610.285 07

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Sistem Komputer
Komputer dalam Ilmu Kedokteran

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment