Image Noise Reduction Based on a Fixed Wavelet Frame and CNNs Applied to CT
Autor: | Luis Albert Zavala-Mondragon, Peter H. N. de With, Fons van der Sommen |
---|---|
Přispěvatelé: | Video Coding & Architectures, Center for Care & Cure Technology Eindhoven, Eindhoven MedTech Innovation Center, EAISI Health |
Rok vydání: | 2021 |
Předmět: |
Image quality
Computer science Noise reduction 02 engineering and technology Iterative reconstruction Signal-To-Noise Ratio wavelets Convolutional neural network Wavelet denoising 0202 electrical engineering electronic engineering information engineering Image Processing Computer-Assisted Humans Signal processing business.industry Deep learning encoder-decoder Pattern recognition Computer Graphics and Computer-Aided Design Haar wavelet shrinkage 020201 artificial intelligence & image processing Artificial intelligence Neural Networks Computer business Tomography X-Ray Computed CNN Software Algorithms |
Zdroj: | IEEE Transactions on Image Processing, 30, 9386-9401. Institute of Electrical and Electronics Engineers |
ISSN: | 1941-0042 1057-7149 |
Popis: | Radiation exposure in CT imaging leads to increased patient risk. This motivates the pursuit of reduced-dose scanning protocols, in which noise reduction processing is indispensable to warrant clinically acceptable image quality. Convolutional Neural Networks (CNNs) have received significant attention as an alternative for conventional noise reduction and are able to achieve state-of-the art results. However, the internal signal processing in such networks is often unknown, leading to sub-optimal network architectures. The need for better signal preservation and more transparency motivates the use of Wavelet Shrinkage Networks (WSNs), in which the Encoding-Decoding (ED) path is the fixed wavelet frame known as Overcomplete Haar Wavelet Transform (OHWT) and the noise reduction stage is data-driven. In this work, we considerably extend the WSN framework by focusing on three main improvements. First, we simplify the computation of the OHWT that can be easily reproduced. Second, we update the architecture of the shrinkage stage by further incorporating knowledge of conventional wavelet shrinkage methods. Finally, we extensively test its performance and generalization, by comparing it with the RED and FBPConvNet CNNs. Our results show that the proposed architecture achieves similar performance to the reference in terms of MSSIM (0.667, 0.662 and 0.657 for DHSN2, FBPConvNet and RED, respectively) and achieves excellent quality when visualizing patches of clinically important structures. Furthermore, we demonstrate the enhanced generalization and further advantages of the signal flow, by showing two additional potential applications, in which the new DHSN2 is used as regularizer: (1) iterative reconstruction and (2) ground-truth free training of the proposed noise reduction architecture. The presented results prove that the tight integration of signal processing and deep learning leads to simpler models with improved generalization. |
Databáze: | OpenAIRE |
Externí odkaz: |