Audy, Shabina Retalia (2025) Deteksi Karies Gigi Berdasarkan Citra Panoramik Menggunakan Convolutiona Network. Other thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
5026211103-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy |
Abstract
Karies gigi merupakan permasalahan kesehatan yang sering terjadi di Indonesia, dengan lebih dari 56% masyarakat mengalami masalah gigi namun hanya 11,2% yang mendapat perawatan medis. Hal ini mendorong dilakukannya pengerjaan Tugas Akhir yang bertujuan untuk mengembangkan sistem deteksi karies gigi otomatis menggunakan Convolutional Neural Network (CNN) pada citra panoramik untuk membantu diagnosis awal. Data yang digunakan adalah 589 citra panoramik gigi format JPEG dari Paraguay. Metodologi dari pengerjaan Tugas Akhir ini meliputi preprocessing citra, pemotongan per gigi, pelabelan karies/sehat, pembagian data training-testing, pengembangan model, serta pembuatan prediksi. Pengerjaan Tugas Akhir ini mengeksplorasi empat pendekatan pembangunan model yaitu dengan dataset gigi individual, citra panoramik utuh, fine-tuning, dan dataset campuran. Hasil menunjukkan model dengan dataset gigi individual memberikan performa terbaik dengan akurasi 81%. Sedangkan untuk model citra panoramik utuh menunjukkan akurasi 61%, fine- tuning mencapai 65%, dan dataset campuran 74%. Pada hasil prediksi aktual, dataset campuran menunjukkan kemampuan deteksi karies terbaik dengan tingkat deteksi 20% dibandingkan pendekatan lainnya.
====================================================================================================================================
Dental caries is a common health problem in Indonesia, with more than 56% of people experiencing dental problems but only 11.2% receiving medical treatment. This study aims to develop an automatic dental caries detection system using Convolutional Neural Network (CNN) on panoramic images for early diagnosis assistance. The research data used 589 JPEG format dental panoramic images from Paraguay. The methodology of this study includes image preprocessing, tooth-by-tooth cropping, caries/healthy labeling, training-testing data sharing, model development, and prediction generation. This study explored four model building approaches: with individual tooth datasets, whole panoramic images, fine-tuning, and mixed datasets. The results showed that the model with individual tooth dataset gave the best performance with 81% accuracy. Meanwhile, the whole panoramic image model showed an accuracy of 61%, fine-tuning reached 65%, and the mixed dataset 74%. In actual prediction results, the mixed dataset showed the best caries detection capability with a detection rate of 20% compared to other approaches.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Deteksi Dini, Karies gigi, Citra Panoramik, Convolutional Neural Network, Image Classification, Early Detection, Dental caries, Panoramic image |
Subjects: | R Medicine > RK Dentistry |
Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis |
Depositing User: | Shabina Retalia Audy |
Date Deposited: | 28 Jul 2025 05:05 |
Last Modified: | 28 Jul 2025 05:05 |
URI: | http://repository.its.ac.id/id/eprint/122607 |
Actions (login required)
![]() |
View Item |