Learning deconvolutional deep neural network for high resolution medical image reconstruction

Liu, Hui; Xu, Jun; Wu, Yan; Guo, Qiang; Ibragimov, Bulat; Li, Xing

doi:10.1016/j.ins.2018.08.022

Cited by 68 publications

(32 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CNN is designed to connect only center and surrounding neurons to enhance feature localizations and learning efficiency . It hence provides superior results for problems in which information required for inference is localized like image deblurring, super‐resolution, and denoising . Scatter effects generally appear in a wide area of images and are difficult to be inferred using a small area in image domain.…”

Section: Discussionmentioning

confidence: 99%

“…Through an appropriate training by using large numbers of paired projection and scatter images, the approach is able to provide a powerful nonlinear predictive model of scatter distribution. Inspired by the superior performance of convolutional neural network (CNN) in dealing with multidimensional data, the approach has been widely used for image classification, segmentation, super‐resolution, and denoising . Specific to medical imaging, various CNN models have been developed for low‐dose fan‐beam CT image restoration, MR‐to‐CT image symthesis, PET image segmentation, and so on .…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Projection‐domain scatter correction for cone beam computed tomography using a residual convolutional neural network

Nomura

Shirato

et al. 2019

Medical Physics

Self Cite

View full text Add to dashboard Cite

images was 0.9993 compared to 0.9984 for the non-tuned-CNN-corrected images or 0.9990 for the fASKS-corrected images. Finally, the CNN-based method is computationally efficientthe correction time for the 360 projections only took less than 5 s in the reported experiments on a PC (4.20 GHz Intel Core-i7 CPU) with a single NVIDIA GTX 1070 GPU. Conclusions: The proposed deep learning-based method provides an effective tool for CBCT scatter correction and holds significant value for quantitative imaging and image-guided radiation therapy.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Projection‐domain scatter correction for cone beam computed tomography using a residual convolutional neural network

Nomura

Shirato

et al. 2019

Medical Physics

Self Cite

View full text Add to dashboard Cite

show abstract

“…Therefore, a dataset for the prototyping can be selected so that it would allow building an SSN (including training, validation, testing) as fast as possible. Obviously, a connection between the dataset and the real SIS task must exist [3], [4], [6].…”

Section: Background and Motivation For Toy Datasetsmentioning

confidence: 99%

“…A downsampling subnetwork is stacked of convolutional layers (ConvLs), ReLUs, and max pooling layers. The upsampling is executed using the transposed convolutional layer, which is also commonly referred to as deconvolutional layer (DeConvL) [4], [6]. DeConvL simultaneously performs the upsampling and filtering, so the upsampling subnetwork is stacked of DeConvLs and ReLUs.…”

Section: Introduction To Semantic Image Segmentationmentioning

confidence: 99%

Generator of a Toy Dataset of Multi-Polygon Monochrome Images for Rapidly Testing and Prototyping Semantic Image Segmentation Networks

Romanuke

2019

Electrical, Control and Communication Engineering

View full text Add to dashboard Cite

In the paper, the problem of building semantic image segmentation networks in a more efficient way is considered. Building a network capable of successfully segmenting real-world images does not require a real semantic image segmentation task. At this stage, called prototyping, a toy dataset can be used. Such a dataset can be artificial and thus may not need augmentation for training. Besides, its entries are images of much smaller size, which allows training and testing the network a way faster. Objects to be segmented are one or few convex polygons in one image. Thus, a toy dataset generator is created whose complexity is regulated by the number of edges in a polygon, the maximal number of polygons in one image, the set of scale factors, and the set of probabilities determining how many polygons in a current image are generated. The dataset capacity and image size are concurrently adjustable, although they are much less influential.

show abstract

“…Biomedical event extraction can help biomedical scientists to do research conveniently, and provide inspiration and basis for the diagnosis, prevention, treatment and new drug research. Also, there are many useful applications for biomedical event task, such as domain search engine [1], pathway curtain [2], medical research [3] and so on. Meanwhile, many evaluation tasks have been organized for providing novel methods of biomedical event extraction tasks, such as BioNLP 2009 [4], BioNLP 2011 [5], BioNLP 2013 [6], and BioNLP 2016 [7].…”

Section: Introductionmentioning

confidence: 99%

A Two-Stage Biomedical Event Trigger Detection Method Based on Hybrid Neural Network and Sentence Embeddings

Ren²,

Tai³

et al. 2021

IEEE Access

View full text Add to dashboard Cite

Biomedical event extraction is a challenging task in biomedical text mining, which plays an important role in improving biomedical research and disease prevention. As the crucial and prerequisite step in event extraction, biomedical trigger detection has attracted much attention. Previous approaches usually depended on feature engineering with unbalanced data. In this paper, we propose a two-stage method based on hybrid neural network for trigger detection, which divides trigger detection into recognition stage and classification stage. In the first stage, we build a BiLSTM based recognition model integrating attention mechanism (Att-BiLSTM). In the second stage, the classification model based on Passive-Aggressive online algorithm is constructed. Furthermore, to enrich sentence-level features, we establish sentence embeddings and add reading gate. On the multi-level event extraction (MLEE) corpus test dataset, our method achieves an F-score of 80.26%, which achieves the state-of-the-art systems.

show abstract

Learning deconvolutional deep neural network for high resolution medical image reconstruction

Cited by 68 publications

References 34 publications

Projection‐domain scatter correction for cone beam computed tomography using a residual convolutional neural network

Projection‐domain scatter correction for cone beam computed tomography using a residual convolutional neural network

Generator of a Toy Dataset of Multi-Polygon Monochrome Images for Rapidly Testing and Prototyping Semantic Image Segmentation Networks

A Two-Stage Biomedical Event Trigger Detection Method Based on Hybrid Neural Network and Sentence Embeddings

Contact Info

Product

Resources

About