Deep learning has achieved a wide range of successful applications in the field of low-dose computed tomography(LDCT) denoising. To balance the relationship between LDCT image denoising and texture details during deep learning training, an LDCT denoising model based on multi-feature extraction and attention mechanism is proposed. The model consists of three branches. The first branch is the shallow edge feature extraction branch, so that the input LDCT image can be fully pre-trained. The second branch uses cross-convolution to explore more edge details in LDCT images. The third branch uses trainable Sobel and normal-dose CT edge labels to further enhance the performance of LDCT image edge details. Finally, the multi-feature fusion module is realized by using the attention mechanism. The experimental results on the American Association of Physicists in Medicine (AAPM) public dataset show that compared with the existing methods, the peak signal-to-noise ratio value of the CT image of the AAPM dataset processed by the new model is 33.637 6, and the structural similarity value is 0.916 9. This method can effectively remove noise and artifacts, while effectively retaining the structural information of the CT image.
WANGYaopeng, JIALina, LIUZhuo. A fast low-dose CT images denosing algorithm based on edge extraction and progressive non-local means[J]. Journal of Test and Measurement Technology, 2023, 37(3): 243-248. (in Chinese)
LIUZhuo, JIALina, WANGYaopeng. Two-step denoising method for low-dose CT images based on low-rank matrix approximation[J]. Journal of Test and Measurement Technology, 2023, 37(5): 449-454. (in Chinese)
[5]
DABOVK, FOI A, KATKOVNIKV, et al. Image denoising by sparse 3-D transform-domain collaborative filtering[J]. IEEE Transactions on Image Processing, 2007, 16(8): 2080-2095.
[6]
HUANGJ, MAJ, LIUN, et al. Projection data restoration guided non-local means for low-dose computed tomography reconstruction[C]//2011 IEEE International Symposium on Biomedical Imaging: From Nano to Macro, 2011: 1167-1170.
[7]
KANGD, SLOMKAP, NAKAZATOR, et al. Image denoising of low-radiation dose coronary CT angiography by an adaptive block-matching 3D algorithm[C]//Medical Imaging: Image Processing, 2013: 86692.
[8]
BUADESA, COLLB, MORELJ M. A non-local algorithm for image denoising[C]//2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05), 2005: 60-65.
[9]
CHENY, YANGZ, HUY, et al. Thoracic low-dose CT image processing using an artifact suppressed large-scale nonlocal means[J]. Physics in Medicine and Biology, 2012, 57(9): 2667-2688.
[10]
FANF, SHANH, KALRAMK, et al. Quadratic autoencoder (Q-AE) for low-dose CT denoising[J]. IEEE Transactions on Medical Imaging, 2019, 39(6): 2035-2050.
[11]
KANGE, MINJ, YEJ C. A deep convolutional neural network using directional wavelets for low-dose X-ray CT reconstruction[J]. Medical Physics, 2017, 44(10): 360-375.
WOLTERINKJ M, LEINERT, VIERGEVERM A, et al. Generative adversarial networks for noise reduction in low-dose CT[J]. IEEE Transactions on Medical Imaging, 2017, 36(12): 2536-2545.
[15]
GENGM, MENGX, YUJ, et al. Content-noise complementary learning for medical image denoising[J]. IEEE Transactions on Medical Imaging, 2022, 41(2): 407-419.
[16]
HANZ, SHANGGUANH, ZHANGX, et al. A dual-encoder-single-decoder based low-dose CT denoising network[J]. IEEE Journal of Biomedical and Health Informatics, 2022, 26(7): 3251-3260.
[17]
WANGD, FANF, WUZ, et al. CTformer: convolution-free Token2Token dilated vision transformer for low-dose CT denoising[J/OL].Electrical Engineering and System Science, 2022. arXiv: 2202.13517.
[18]
LIANGT, JINY, LIY, et al. EDCNN: edge enhancement-based densely connected network with compound loss for low-dose CT denoising[C]//2020 15th IEEE International Conference on Signal Processing (ICSP), 2020: 193-198.
[19]
AAPM. Low dose CT grand challenge[EB/OL]. [2024-08-25].
[20]
KAZAKOVAN, MARGALAM, DURDLEN G. Sobel edge detection processor for a real-time volume rendering system[C]//2004 IEEE International Symposium on Circuits and Systems, 2004: 913.
[21]
FANGF, LIJ, YUANY, et al. Multilevel edge features guided network for image denoising[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020, 32(9): 3956-3970.
[22]
KANGE, MINJ, YEJ C. Wavelet domain residual network (WavResNet) for low-dose X-ray CT reconstruction[J/OL]. Computer Science, 2017. arXiv: 1703.01383.
[23]
ISOLAP, ZHUJ Y, ZHOUT, et al. Image-to-image translation with conditional adversarial networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017: 5967-5976.
[24]
WANGZ, BOVIKA C, SHEIKHH R, et al. Image quality assessment: from error visibility to structural similarity[J]. IEEE Transactions on Image Processing, 2004, 13(4): 600-612.
[25]
XUEW, ZHANGL, MOUX, et al. Gradient magnitude similarity deviation: a highly efficient perceptual image quality index[J]. IEEE Transactions on Image Processing, 2013, 23(2): 684-695.
[26]
ZHANGL, ZHANGL, MOUX, et al. FSIM: a feature similarity index for image quality assessment[J]. IEEE Transactions on Image Processing, 2011, 20(8): 2378-2386.
[27]
SHEIKHH R, BOVIKA C. Image information and visual quality[J]. IEEE Transactions on Image Processing, 2006, 15(2): 430-444.