MS-DRDNet: Optimization-Inspired Deep Compressive Sensing Network for MRI

Yue, Huihui; Guo, Jun; Yin, Xiangjun

doi:10.1007/978-3-030-60636-7_32

Cited by 2 publications

(3 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, Zhang et al propose an OPINE-Net and an enhanced version OPINE-Net + to realize image compression and reconstruction end-to-end [19]. MS-DRDNet [22] achieves CS compression and reconstruction by jointly optimizing its four components. In summary, these studies enrich the development of deep CS network in image recovery.…”

Section: End-to-end Learning Methodsmentioning

confidence: 99%

“…In this section, we show the image reconstruction performance of the proposed MB-DAMPNet by comparing with the state-of-the-art BM3D-IT [12], BM3D-AMP [13], DeepInverse [20], ReconNet [21], OPINE-Net + [19], MS-DRDNet [22], and SCSNet [24] methods. To ensure the fairness of the experiments, all the data-driven deep learning methods employed for comparison (the last five methods) use the same training dataset as the proposed MB-DAMPNet.…”

Section: Comparison With State-of-the-art Methodsmentioning

confidence: 99%

“…OPINE-Net [19] is an advanced version of ISTA-Net, which can be used to reconstruct image end-to-end. Moreover, DeepInverse [20], ReconNet [21], MS-DRDNet [22], CSNet [23], and SCSNet [24] are also effective deep network structures designed by imitating the theory of CS. However, the common problem of these methods is that the size of measurement matrix A is fixed or hand-crafted, so all of them can only reconstruct the input image with a specific size.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

MB-DAMPNet: a novel multi-branch denoising-based approximate message passing algorithm via deep neural network for image reconstruction

Yue¹,

Guo²,

Yin³

et al. 2021

Inverse Problems

View full text Add to dashboard Cite

Compressive sensing has gained great attention due to its effectiveness in solving linear inverse problems. However, how to further improve the accuracy of compressed image inversion while ensuring or even accelerating the speed is still a major challenge. To tackle this problem, we present a novel multi-branch denoising-based approximate message passing algorithm via deep neural network, dubbed MB-DAMPNet. It mainly consists of three components, i.e. sampling subnet, initial recovery subnet, and deep recovery subnet, which are optimized jointly. The sampling subnet is constructed to obtain the compressed measurements, the initial recovery subnet is employed to generate the reconstructed image by inverse transformation, while the deep recovery subnet is designed to refine the reconstructed results obtained by the former, so as to improve the image accuracy. Moreover, the matrix multiplication in the network is all designed as matrix convolution which can be learned automatically, so that the input image of the MB-DAMPNet can be of different scales, which improves the flexibility and applicability of the network. In addition, all parameters in the network are learned end-to-end instead of fixed or hand-crafted. The numerical results validate that our method significantly outperforms other state-of-the-art methods in image reconstruction accuracy.

show abstract

Section: End-to-end Learning Methodsmentioning

confidence: 99%