SepUnet: Depthwise Separable Convolution Integrated U-Net For MRI Reconstruction

Zabihi, Soheil; Rahimian, Elahe; Asif, Amir; Mohammadi, Arash

doi:10.1109/icip42928.2021.9506285

Cited by 7 publications

(3 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The architecture includes a compressed path for acquiring information and an extended path for performing precise localization. SepUNet [ 56 ] proposes a model that significantly reduces the number of parameters required while maintaining a high level of accuracy and can effectively reduce the computational cost associated with single-coil reconstruction tasks. MSFCN [ 57 ] introduces a multisupervised side output layer in a deep end-to-end network to guide multiscale feature training.…”

Section: Simulation Analysismentioning

confidence: 99%

See 1 more Smart Citation

A Residual Fusion Network for Osteosarcoma MRI Image Segmentation in Developing Countries

Zhou

Gou

et al. 2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

Among primary bone cancers, osteosarcoma is the most common, peaking between the ages of a child’s rapid bone growth and adolescence. The diagnosis of osteosarcoma requires observing the radiological appearance of the infected bones. A common approach is MRI, but the manual diagnosis of MRI images is prone to observer bias and inaccuracy and is rather time consuming. The MRI images of osteosarcoma contain semantic messages in several different resolutions, which are often ignored by current segmentation techniques, leading to low generalizability and accuracy. In the meantime, the boundaries between osteosarcoma and bones or other tissues are sometimes too ambiguous to separate, making it a challenging job for inexperienced doctors to draw a line between them. In this paper, we propose using a multiscale residual fusion network to handle the MRI images. We placed a novel subnetwork after the encoders to exchange information between the feature maps of different resolutions, to fuse the information they contain. The outputs are then directed to both the decoders and a shape flow block, used for improving the spatial accuracy of the segmentation map. We tested over 80,000 osteosarcoma MRI images from the PET-CT center of a well-known hospital in China. Our approach can significantly improve the effectiveness of the semantic segmentation of osteosarcoma images. Our method has higher F1, DSC, and IOU compared with other models while maintaining the number of parameters and FLOPS.

show abstract

Section: Simulation Analysismentioning

confidence: 99%

“…SepUNet [ 56 ] proposes a model that significantly reduces the number of parameters required while maintaining a high level of accuracy and can effectively reduce the computational cost associated with single-coil reconstruction tasks.…”

Section: Simulation Analysismentioning

confidence: 99%

A Residual Fusion Network for Osteosarcoma MRI Image Segmentation in Developing Countries

Zhou

Gou

et al. 2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

show abstract

“…Moreover, depthwise separable convolution also has similar applications on the task of image reconstruction. For example, Zabihi et al (2021) integrated depthwise separable convolution and Atrous Spatial Pyramid Pooling (ASPP) module into Unet, which not only improved the reconstruction accuracy of magnetic resonance imaging (MRI) but also reduced the number of parameters and memory consumption by the network. The aforementioned methods using depthwise separable convolution are two-dimensional, while in geophysics, many seismic data are three-or five-dimensional.…”

Section: Introductionmentioning

confidence: 99%

Depthwise separable convolution Unet for 3D seismic data interpolation

Jin

Yang

et al. 2023

Front. Earth Sci.

View full text Add to dashboard Cite

In seismic exploration, dense and evenly spatial sampled seismic traces are crucial for successful implementation of most seismic data processing and interpretation algorithms. Recently, numerous seismic data reconstruction approaches based on deep learning have been presented. High dimension-based methods have the benefit of making full use of seismic signal at different perspectives. However, with the transformation of data dimension from low to high, the parameter capacity and computation cost of training deep neural network increase significantly. In this paper, we introduce depthwise separable convolution instead of standard convolution to reduce the operation cost of Unet for 3D seismic data missing trace interpolation. The structural similarity (SSIM), L1 hybrid loss function, and switchable normalization further improve the reconstruction performance of the network. The comparative experiments on the synthetic and field seismic data show that depthwise separable convolution can effectively reduce the number of network parameters and computation intensity with the interpolation results comparable to the standard convolution results.

show abstract

Transformer‐ and joint learning‐based dual‐domain networks for undersampled MRI segmentation

Duan,

Huang,

Xie

et al. 2024

Medical Physics

View full text Add to dashboard Cite

BackgroundRecently, magnetic resonance imaging (MRI) has become a crucial medical imaging technology widely used in clinical practice. However, MRI faces challenges such as the lengthy acquisition time for k‐space data and the need for time‐consuming manual annotation by radiologists. Traditionally, these challenges have been addressed individually through undersampled MRI reconstruction and automatic segmentation algorithms. Whether undersampled MRI segmentation can be enhanced by treating undersampled MRI reconstruction and segmentation as an end‐to‐end task, trained simultaneously, rather than as serial tasks should be explored.PurposeWe introduce a novel Transformer‐ and Joint Learning‐based Dual‐domain Network (TJLD‐Net) for undersampled MRI segmentation.MethodsThis method significantly enhances feature recognition in the segmentation process by fully utilizing the rich detail obtained during the image reconstruction phase. Consequently, the method can achieve precise and reliable image segmentation even with undersampled k‐space data. Additionally, it incorporates an attention mechanism for feature enhancement, which improves the representation of shared features by learning the contextual information in MR images.ResultsSimulation experiments demonstrate that the segmentation performance of TJLD‐Net on three datasets is significantly higher than that of the joint model (RecSeg) and six baseline models (where reconstruction and segmentation are regarded as serial tasks). On the CHAOS dataset, the Dice scores of TJLD‐Net are, on average, 9.87%, 2.17%, 1.90%, 1.80%, 9.60%, 0.80%, and 6.50% higher than those of the seven compared models. On the ATLAS challenge dataset, the average Dice scores of TJLD‐Net improve by 4.23%, 5.63%, 2.30%, 1.53%, 3.57%, 0.93%, and 6.60%. Similarly, on the SKM‐TEA dataset, the average Dice scores of TJLD‐Net improve by 4.73%, 12.80%, 14.83%, 8.67%, 4.53%, 11.60%, and 12.10%. The novel TJLD‐Net model provides a promising solution for undersampled MRI segmentation, overcoming the poor performance issues encountered by automated segmentation algorithms in low‐quality accelerated imaging.

show abstract

SepUnet: Depthwise Separable Convolution Integrated U-Net For MRI Reconstruction

Cited by 7 publications

References 21 publications

A Residual Fusion Network for Osteosarcoma MRI Image Segmentation in Developing Countries

A Residual Fusion Network for Osteosarcoma MRI Image Segmentation in Developing Countries

Depthwise separable convolution Unet for 3D seismic data interpolation

Transformer‐ and joint learning‐based dual‐domain networks for undersampled MRI segmentation

Contact Info

Product

Resources

About