Image of Penerapan Algoritma Faster R-Cnn Dalam Sistem Deteksi Karies Gigi Pada Citra X-Ray
Penanda Bagikan

CD Skripsi

Penerapan Algoritma Faster R-Cnn Dalam Sistem Deteksi Karies Gigi Pada Citra X-Ray

ABSTRACT
Manual Manual dental caries detection using X-ray images has limitations related to time, subjectivity, and the potential for interpretation errors, especially in early-stage caries cases. This research aims to evaluate and compare various configurations of the Faster R-CNN algorithm to identify the most accurate automatic dental caries detection system for panoramic X-ray images. An experimental research method was employed, where several Faster R-CNN models with a ResNet-50 and FPN backbone were designed and tested by manipulating variables such as the CBAM attention module, a custom AnchorGenerator, and CustomRoIHeads with Focal Loss. Training results over 20 epochs showed a best validation mAP performance of 0.9094. Evaluation on an independent test set yielded a Test mAP of 0.8382, Test Precision of 0.6396, and Test Recall of 0.7585. Testing on 25 test data samples showed a Precision of 83.93%, Recall of 88.68%, and an F1 Score of 86.24%. The best-performing model was implemented in a Graphical User Interface (GUI), “Dental Decay Detection Studio,” using Gradio to facilitate its use by dentists. The experimental results indicate that a specific Faster R-CNN configuration can build a promising dental caries detection system, potentially enhancing clinical diagnostic accuracy and serving as an effective auxiliary tool for medical practitioners.
Keywords: Dental Caries Detection, Faster R-CNN, Panoramic X-ray Images, Deep Learning, Artificial Intelligence

Ketersediaan
#
Perpustakaan Universitas Riau 2107125642
2107125642
Tersedia
Informasi Detail
Judul Seri
-
No. Panggil
2107125642
Penerbit
Pekanbaru : Universitas Riau – F.TEKNIK – INFORMATIKA.,
Deskripsi Fisik
-
Bahasa
Indonesia
ISBN/ISSN
-
Klasifikasi
2107125642
Tipe Isi
-
Tipe Media
-
Tipe Pembawa
-
Edisi
-
Subjek
TEKNIK INFORMATIKA
Info Detail Spesifik
-
Pernyataan Tanggungjawab
Versi lain/terkait

Tidak tersedia versi lain

Lampiran Berkas
Komentar
Anda harus masuk sebelum memberikan komentar