Related Experiment Video
Updated: Nov 27, 2025
09:10
Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures
Published on: August 5, 2021
2.1K
Using super-resolution generative adversarial network models and transfer learning to obtain high resolution digital
Maira B H Moran1, Marcelo D B Faria2, Gilson A Giraldi3
1Policlínica Piquet Carneiro, Universidade Do Estado Do Rio de Janeiro, 20950-003, Rio de Janeiro, Brazil; Instituto de Computação, Universidade Federal Fluminense, 24210-310, Niterói, Brazil.
Computers in Biology and Medicine
|December 3, 2020
Summary
Super-resolution generative adversarial networks (SRGAN) with transfer learning enhance periapical radiograph quality. This deep learning approach improves diagnostic accuracy by generating higher-resolution images compared to traditional methods.
Area of Science:
- Dentistry
- Medical Imaging
- Artificial Intelligence
Background:
- Periapical radiographs are crucial for diagnosing dental diseases.
- Image quality limitations can hinder accurate diagnosis.
- Current commercial super-resolution methods may introduce artifacts.
Purpose of the Study:
- To evaluate super-resolution generative adversarial networks (SRGAN) with transfer learning for enhancing periapical radiograph quality.
- To compare SRGAN performance against traditional super-resolution techniques.
- To assess the impact of transfer learning datasets on image enhancement.
Main Methods:
- Implementation of SRGAN models with and without transfer learning.
- Comparison with interpolation-based super-resolution methods.
- Quantitative evaluation using MSE, PSNR, SSIM, and MOS metrics.
- Statistical analysis using the Wilcoxon paired test.
Main Results:
- SRGAN models utilizing transfer learning demonstrated superior performance on average across all metrics.
- Statistical analysis confirmed the significant superiority of transfer learning approaches.
- Visual analysis revealed enhanced edge details and reduced blur effects in SRGAN-generated images.
Conclusions:
- Transfer learning significantly improves SRGAN performance for periapical radiograph super-resolution.
- SRGAN offers a promising deep learning solution for improving dental image quality and diagnostic accuracy.
- The study highlights the potential of AI in overcoming limitations of conventional dental imaging techniques.
Keywords:
Generative adversarial networksImage enhancementPeriapical radiographySuper-resolutionTransfer learning
