Background
As an imaging modality, cone beam CT (CBCT) is widely employed in dentistry, which can help dentists to observe tissues such as roots and jaws without confusion. CBCT has the advantages of convenience and low radiation dose, however low contrast and large noise are the critical issues in the images. These disadvantages make it difficult for dentists to identify target tissues accurately. Due to the differences in scanning methods and reconstruction algorithms between CBCT and multi-row detector spiral CT (MDCT), the current CT noise reduction models present significant shortcomings in terms of reducing the noise on CBCT images.
Results
In this paper, we propose a method of image noise reduction based on a conditional generative adversarial network to improve the quality of CBCT images. The normal-dose MDCT images are used as the ground truth images for model training to generate the denoised images. In order to increase the model’s sensitivity to the gradient information, a gradient loss function is involved in our proposed method. The verification experiments on the simulated data set and the real data set show that our model effectively generates denoised images as well as preserves the quality of the images. We compare the denoising effect between our model and other models with different loss functions. The scores by PSNR, MS-SSIM and GMSD show that our model presents better edge characteristics and a denoising effect.
Conclusion
Through the experiments, the performance of our improved cGAN for denoising in oral CBCT data are better than the compared with other models in terms of evaluation score and visual quality. Our model is more competitive for clinical application. Comparing with other models which only reduce simulated noise, we directly input CBCT images to reduce the noise.