Low-Dose CT Denoising Using A Structure-Preserving Kernel Prediction Network

• Published in: 2021 IEEE International Conference on Image Processing (ICIP) pp. 1639 - 1643
• Main Authors: Xu, Lu, Zhang, Yuwei, Liu, Ying, Wang, Daoye, Zhou, Mu, Ren, Jimmy, Wei, Jingwei, Ye, Zhaoxiang
• Format: Conference Proceeding
• Language: English
• Published: IEEE 19.09.2021
• Subjects: Computed tomography; Conferences; Image Denoising; Image processing; Imaging; Kernel Prediction Network; Low-dose CT; Noise reduction; Predictive models; Protocols
• ISSN: 2381-8549
• DOI: 10.1109/ICIP42928.2021.9506679

Low-dose CT has been a key diagnostic imaging modality to reduce the potential risk of radiation overdose to patient health. Despite recent advances, CNN-based approaches typically apply filters in a spatially invariant way and adopt similar pixel-level losses, which treat all regions of the CT image equally and can be inefficient when fine-grained structures coexist with non-uniformly distributed noises. To address this issue, we propose a Structure-preserving Kernel Prediction Network (StructKPN) that combines the kernel prediction network with a structure-aware loss function that utilizes the pixel gradient statistics and guides the model towards spatially-variant filters that enhance noise removal, prevent over-smoothing and preserve detailed structures for different regions in CT imaging. Extensive experiments demonstrated that our approach achieved superior performance on both synthetic and non-synthetic datasets, and better preserves structures that are highly desired in clinical screening and low-dose protocol optimization.