Deep denoising for multi-dimensional synchrotron X-ray tomography without high-quality reference data

Hendriksen, Allard A.; Bührer, Minna; Leone, Laura; Merlini, Marco; Viganò, Nicola; Pelt, Daniël M.; Marone, Federica; Michiel, Marco Di; Batenburg, Kees Joost

doi:10.1038/s41598-021-91084-8

Cited by 33 publications

(21 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In practice, this means that large images can be efficiently processed without running out of computer memory, and accurate training is possible with a limited amount of training data. These advantages have already proven effective for various non-cycloidal applications of CT in earlier work 17 , 25 – 27 , and make the MS-D CNN especially applicable to the task of data recovery in cycloidal CT. Recently, other CNN architectures were proposed that also make use of dilated convolutions and dense connections to capture multi-scale information 32 – 34 .…”

Section: Methodsmentioning

confidence: 93%

“…In this paper, we report on applying machine learning to the sinogram completion task in cycloidal CT, with the aim of reconstructing high quality images from incomplete (low-dose) data, and demonstrating that bicubic interpolation (which was applied previously) can be outperformed. Over recent years, machine learning has seen a surge in application in the context of (standard) CT 15 – 27 . Convolutional neural networks (CNNs) 28 – 34 have been applied as part of the tomographic image reconstruction process 18 , thereby improving the reconstruction quality, or as a post-processing tool to improve the quality of CT images after they have been reconstructed 15 – 17 .…”

Section: Introductionmentioning

confidence: 99%

“…In many cases, the acquisition of such training data can be either impossible or impractical, because it may be time-consuming or labour intensive, both in experimental and processing time, or the sample may be unique or dose-sensitive. In recent years, self-supervised learning, an approach that does not require any additional reference data, has shown promising results in tomographic reconstruction 26 , 27 . However, these methods are typically only applicable to the denoising of reconstructed images and rely on specific mathematical assumptions about the acquired data which makes them not directly usable for the sinogram completion task in cycloidal CT.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Cycloidal CT with CNN-based sinogram completion and in-scan generation of training data

Pelt

Morgó

Jones

et al. 2022

Sci Rep

Self Cite

View full text Add to dashboard Cite

In x-ray computed tomography (CT), the achievable image resolution is typically limited by several pre-fixed characteristics of the x-ray source and detector. Structuring the x-ray beam using a mask with alternating opaque and transmitting septa can overcome this limit. However, the use of a mask imposes an undersampling problem: to obtain complete datasets, significant lateral sample stepping is needed in addition to the sample rotation, resulting in high x-ray doses and long acquisition times. Cycloidal CT, an alternative scanning scheme by which the sample is rotated and translated simultaneously, can provide high aperture-driven resolution without sample stepping, resulting in a lower radiation dose and faster scans. However, cycloidal sinograms are incomplete and must be restored before tomographic images can be computed. In this work, we demonstrate that high-quality images can be reconstructed by applying the recently proposed Mixed Scale Dense (MS-D) convolutional neural network (CNN) to this task. We also propose a novel training approach by which training data are acquired as part of each scan, thus removing the need for large sets of pre-existing reference data, the acquisition of which is often not practicable or possible. We present results for both simulated datasets and real-world data, showing that the combination of cycloidal CT and machine learning-based data recovery can lead to accurate high-resolution images at a limited dose.

show abstract

Section: Methodsmentioning

confidence: 93%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Cycloidal CT with CNN-based sinogram completion and in-scan generation of training data

Pelt

Morgó

Jones

et al. 2022

Sci Rep

Self Cite

View full text Add to dashboard Cite

show abstract

“… – 27 This 2.5D CNN can exploit 3D geometric information while still using a 2D CNN architecture. Although some researchers stated that the 3D structures may be lost by the first 2D convolution in the 2.5D CNN model, 28 others believed that the 2.5D CNN can often achieve an image quality similar to 3D CNNs at a reduced computational cost 26 , 27 , 29 . In this work, we did not refer to this model as “2.5D CNN” because this name was also used to represent a CNN model that uses both 2D and 3D convolutions in the same model, 30 which is similar to our ResNet3D-v1 model.…”

Section: Discussionmentioning

confidence: 99%

Multislice input for 2D and 3D residual convolutional neural network noise reduction in CT

et al. 2023

View full text Add to dashboard Cite

Purpose: Deep convolutional neural network (CNN)-based methods are increasingly used for reducing image noise in computed tomography (CT). Current attempts at CNN denoising are based on 2D or 3D CNN models with a single-or multiple-slice input. Our work aims to investigate if the multiple-slice input improves the denoising performance compared with the singleslice input and if a 3D network architecture is better than a 2D version at utilizing the multislice input.Approach: Two categories of network architectures can be used for the multislice input. First, multislice images can be stacked channel-wise as the multichannel input to a 2D CNN model. Second, multislice images can be employed as the 3D volumetric input to a 3D CNN model, in which the 3D convolution layers are adopted. We make performance comparisons among 2D CNN models with one, three, and seven input slices and two versions of 3D CNN models with seven input slices and one or three output slices. Evaluation was performed on liver CT images using three quantitative metrics with full-dose images as reference. Visual assessment was made by an experienced radiologist.Results: When the input channels of the 2D CNN model increases from one to three to seven, a trend of improved performance was observed. Comparing the three models with the seven-slice input, the 3D CNN model with a one-slice output outperforms the other models in terms of noise texture and homogeneity in liver parenchyma as well as subjective visualization of vessels. Conclusions:We conclude the that multislice input is an effective strategy for improving performance for 2D deep CNN denoising models. The pure 3D CNN model tends to have a better performance than the other models in terms of continuity across axial slices, but the difference was not significant compared with the 2D CNN model with the same number of slices as the input.

show abstract

“…A second option is to advance to more efficient temporal filters that are tailored to the physical dynamics and imaging artifacts that are presented in these acquisitions (Mohan et al., 2015). In recent years, deep learning methods have also become a valuable approach to get more results out of challenging μ CT scans (Hendriksen et al., 2021).…”

Section: Discussionmentioning

confidence: 99%

Dynamic CT Reconstruction With Improved Temporal Resolution for Scanning of Fluid Flow in Porous Media

Goethals

Bultreys

Manigrasso

et al. 2022

Water Resources Research

View full text Add to dashboard Cite

Lab‐based high‐resolution Computed Tomography (μCT) is an important tool to investigate multiphase fluid flow through porous media. Rapid and continuous μCT acquisition can be used to resolve the dynamics of the 3D fluid distribution in the pores over time while the flow processes occur. However, the temporal resolution of this technique remains limited due to a trade‐off between acquisition time and image quality. This work presents a method to improve the temporal resolution of multiphase flow imaging experiments by limiting the angular range of radiographs used to reconstruct each time step, while compensating for this loss of information by including a temporal total variation term in the reconstruction algorithm. This addition penalizes improper temporal fluctuations in the reconstructed images, but not those dynamic events that are consistent with the radiographs. We perform a thorough evaluation of the resulting gain in temporal resolution at the single‐pore level. The method is validated on both simulated and experimental data representing multiphase flow in porous media. We find that this method improved the temporal resolution up to a factor 3 compared to reconstructions that use full 360° rotations for each time step.

show abstract

Deep denoising for multi-dimensional synchrotron X-ray tomography without high-quality reference data

Cited by 33 publications

References 47 publications

Cycloidal CT with CNN-based sinogram completion and in-scan generation of training data

Cycloidal CT with CNN-based sinogram completion and in-scan generation of training data

Multislice input for 2D and 3D residual convolutional neural network noise reduction in CT

Dynamic CT Reconstruction With Improved Temporal Resolution for Scanning of Fluid Flow in Porous Media

Contact Info

Product

Resources

About