X-RAY computed tomography (CT) is one of the most valuable imaging techniques in clinics. However, X-ray CT causes potential cancer risks due to radiation exposure. To ensure patient safety, X-ray dose reduction techniques have been extensively studied, and the reduction in the number of X-ray photons using tube current modulation is considered one of the solutions.A drawback of this approach is, however,the low signal-to-noise ratio (SNR) of projections, which induces noise in the reconstructed image. Model based iterative reconstruction (MBIR) algorithms for low-dose X-ray CT are computationally expensive. To address this problem, here a novel framelet-based denoising algorithm is proposed using wavelet residual network which synergistically combines the expressive power of deep learning and the performance guarantee from the framelet-based denoising algorithms. Accordingly, the main goal of this paper is to synergistically combine the expressive power of deep neural network and the performance guarantee from the frameletbased denoising algorithms. In particular, it is shown that the performance of the deep learning-based denoising algorithm can be improved with iterative steps similar to the classical framelet-based denoising approaches. Proposed network is good at streaking noise reduction and preserving the texture details of the organs while the lesion information is maintained.
Input variables : Low-dose X-ray CT images
Output Variables : Clear reconstructed image
Visit Model : github.com
Additional links : arxiv.org
| Model Category | : | Public |
| Date Published | : | March, 2018 |
| Healthcare Domain | : | Medical Technology |
| Code | : | github.com |
| Medical Imaging |
| Image Synthesis |