Sudrajab, Achmad Fajri (2026) Analisis Komparasi Teknik Image Enhancement Untuk Peningkatan Kinerja Deteksi Fraktur Tulang Menggunakan Model YOLOv12. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5025221104-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5025221104-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (16MB) | Request a copy |
Abstract
Fraktur tulang merupakan masalah kesehatan global yang terus meningkat, sementara interpretasi manual citra radiografi sinar-X masih rentan terhadap kesalahan diagnostik akibat
kualitas citra yang rendah. Penelitian ini menganalisis secara komparatif efektivitas tiga teknik image enhancement, yaitu CLAHE, Gamma Correction, dan External Exposure Correction, baik secara tunggal maupun dalam berbagai konfigurasi kombinasi, sebagai tahap prapemrosesan untuk meningkatkan kinerja deteksi dan klasifikasi fraktur tulang menggunakan model YOLOv12 pada dataset FracAtlas. Skenario eksperimen yang diuji mencakup masingmasing teknik tunggal serta kombinasi dari beberapa teknik dan dijalankan secara sistematis dengan konfigurasi pelatihan yang seragam menggunakan model YOLOv12 berbasis transfer learning. Kinerja model dievaluasi pada dua tugas secara terpisah, yaitu tugas deteksi menggunakan metrik mAP@0.5 dan mAP@0.5:0.95, serta tugas klasifikasi level-citra menggunakan Akurasi, F1-Score, dan AUC-ROC. Hasil penelitian menunjukkan bahwa penerapan image enhancement secara umum memberikan pengaruh positif, dengan 88,9% skenario meningkatkan performa klasifikasi dan 55,6% skenario meningkatkan performa deteksi dibandingkan baseline. Konfigurasi terbaik adalah SK-35 dengan pipeline tiga teknik
berurutan, yaitu Gamma Correction (γ=1.2), External Exposure USM Halus, dan CLAHE Konservatif, yang menghasilkan mAP@0.5 sebesar 0,6691 dengan peningkatan 13,5% dari baseline serta Recall deteksi tertinggi sebesar 0,6159, meminimalkan kasus fraktur yang terlewat secara diagnostik.
==================================================================================================================================
Bone fractures represent a growing global health burden, while manual interpretation of X-ray radiographs remains susceptible to diagnostic errors due to poor image quality. This study comparatively analyzes the effectiveness of three image enhancement techniques, namely CLAHE, Gamma Correction, and External Exposure Correction, applied individually and in various combination configurations as a preprocessing stage to improve bone fracture detection
and image-level classification using YOLOv12 on the FracAtlas dataset. The evaluated experimental scenarios encompass both individual techniques and their combinations. These scenarios are executed systematically under a uniform training configuration, utilizing a transfer learning-based YOLOv12 model. Performance was evaluated on two separate tasks: object detection using mAP@0.5 and mAP@0.5:0.95, and image-level classification using Accuracy, F1-Score, and AUC-ROC. Results show that image enhancement generally exerts a positive influence, with 88.9% of scenarios improving classification performance and 55.6% improving
detection performance relative to baseline. The best-performing configuration was SK-35, a three-stage sequential pipeline of Gamma Correction (γ=1.2), Soft Unsharp Masking-based External Exposure Correction, and Conservative CLAHE, achieving a mAP@0.5 of 0.6691 with a 13.5% improvement over baseline and the highest detection Recall of 0.6159, minimizing missed fracture detections in diagnostic contexts.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Image Enhancement, Deteksi Fraktur Tulang, YOLOv12, CLAHE, Gamma Correction, External Exposure Correction, FracAtlas. Image Enhancement, Bone Fracture Detection, YOLOv12, CLAHE, Gamma Correction, External Exposure Correction, FracAtlas. |
| Subjects: | Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QM Human anatomy > QM101 Skeleton. Osteology R Medicine > R Medicine (General) > R858 Deep Learning 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: | ACHMAD FAJRI SUDRAJAB |
| Date Deposited: | 24 Jul 2026 03:49 |
| Last Modified: | 24 Jul 2026 03:49 |
| URI: | http://repository.its.ac.id/id/eprint/136801 |
Actions (login required)
 |
View Item |