Deep learning-based artefact reduction in low-dose dental cone beam computed tomography with high-attenuation...

Area of Science:

• Medical Imaging
• Applied Mathematics
• Computer Science

Background:

• Cone beam CT (CBCT) is increasingly used in dental clinics due to its cost-effectiveness compared to medical CT.
• Low-dose dental CBCT systems face significant image quality degradation challenges.
• Metallic implants, common in dental patients, exacerbate image quality issues due to artifacts.

Purpose of the Study:

• To identify research directions for enhancing image quality in low-dose, cost-effective dental CBCT.
• To critically examine mathematical methodologies for CT image reconstruction.
• To evaluate the potential and limitations of deep learning in addressing CBCT image quality problems.

Main Methods:

• Mathematical analysis of current computed tomography (CT) methodologies.
• Investigation of metal-induced artifacts in CBCT reconstruction.
• Exploration of alternative reconstruction approaches bypassing the traditional Radon transform model.
• Examination of deep learning-based methods for image quality improvement.

Main Results:

• Identified challenges in conventional CT reconstruction methods, particularly concerning metal artifacts.
• Presented an alternative mathematical approach to the Radon transform for improved reconstruction.
• Assessed the efficacy of deep learning techniques in mitigating image quality issues in low-dose dental CBCT.

Conclusions:

• Further research is needed to optimize low-dose dental CBCT image quality, especially in the presence of metal artifacts.
• Alternative mathematical models and deep learning show promise for overcoming current limitations.
• Enhancing dental CBCT image quality is crucial for accurate diagnosis and treatment planning.