Noise2Inverse: Self-Supervised Deep Convolutional Denoising for Tomography

Hendriksen, Allard A.; Pelt, Daniël M.; Batenburg, Kees Joost

doi:10.1109/tci.2020.3019647

Cited by 94 publications

(87 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the interior scanning mode, the X-ray source only radiates about 1/10 of the diameter of the field of view but improvement in image resolution by four times (roughly, from 200 µm to 50 µm) would increase radiation dose significantly (two orders of magnitude) [12]. Thanks to the latest advancement in deep learning based low-dose CT imaging techniques [30], we can reduce radiation dose by an order of magnitude. With all of these factors coupled together, we should be able to maintain approximately the current head CT dose for an interior micro-CT scan to achieve about 50 µm resolution.…”

Section: B Design Principlesmentioning

confidence: 99%

Clinical Micro-CT Empowered by Interior Tomography, Robotic Scanning, and Deep Learning

Zheng

Cong

et al. 2020

IEEE Access

View full text Add to dashboard Cite

While micro-CT systems are instrumental in preclinical research, clinical micro-CT imaging has long been desired with cochlear implantation as a primary application. The structural details of the cochlear implant and the temporal bone require a significantly higher image resolution than that (about 0.2 mm) provided by current medical CT scanners. In this paper, we propose a clinical micro-CT (CMCT) system design integrating conventional spiral cone-beam CT, contemporary interior tomography, deep learning techniques, and the technologies of a micro-focus X-ray source, a photon-counting detector (PCD), and robotic arms for ultrahigh-resolution localized tomography of a freely-selected volume of interest (VOI) at a minimized radiation dose level. The whole system consists of a standard CT scanner for a clinical CT exam and VOI specification, and a robotic micro-CT scanner for a local scan of high spatial and spectral resolution at minimized radiation dose. The prior information from the global scan is also fully utilized for background compensation of the local scan data for accurate and stable VOI reconstruction. Our results and analysis show that the proposed hybrid reconstruction algorithm delivers accurate high-resolution local reconstruction, and is insensitive to the misalignment of the isocenter position, initial view angle and scale mismatch in the data/image registration. These findings demonstrate the feasibility of our system design. We envision that deep learning techniques can be leveraged for optimized imaging performance. With highresolution imaging, high dose efficiency and low system cost synergistically, our proposed CMCT system has great promise in temporal bone imaging as well as various other clinical applications.INDEX TERMS Clinical micro-CT, Deep learning, High-resolution imaging, Interior tomography, Photoncounting detector, Robotic arms, Temporal bone imaging, X-ray computed tomography This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.

show abstract

Section: B Design Principlesmentioning

confidence: 99%

Clinical Micro-CT Empowered by Interior Tomography, Robotic Scanning, and Deep Learning

Zheng

Cong

et al. 2020

IEEE Access

View full text Add to dashboard Cite

show abstract

“…To sidestep the issue of obtaining training data, CNN-based denoising techniques have recently been proposed that do not require the acquisition of high-quality images 17 – 24 . However, many of these techniques rely on assumptions about the source of noise that are not correct in tomography, resulting in suboptimal denoising accuracy 25 . As a solution to these difficulties, we have proposed Noise2Inverse 25 , which is a post-processing technique specifically designed for tomography and related inverse problems.…”

Section: Introductionmentioning

confidence: 99%

“…However, many of these techniques rely on assumptions about the source of noise that are not correct in tomography, resulting in suboptimal denoising accuracy 25 . As a solution to these difficulties, we have proposed Noise2Inverse 25 , which is a post-processing technique specifically designed for tomography and related inverse problems. Using self-supervised training, the acquisition process is exploited to create pairs of noisy reconstructions from a single tomographic dataset.…”

Section: Introductionmentioning

confidence: 99%

“…Using self-supervised training, the acquisition process is exploited to create pairs of noisy reconstructions from a single tomographic dataset. Although no high-quality reconstructions are presented to the network, training still produces a denoising CNN whose accuracy is comparable to supervised CNN training 25 . To obtain a denoised reconstruction, the trained CNN can be applied both to the noisy training reconstruction and to similar data that is acquired in the future.…”

Section: Introductionmentioning

confidence: 99%

“…To obtain a denoising CNN, the reconstructed training pairs must satisfy certain conditions, which we describe in this paper and refer to as the Noise2Inverse conditions . In previous work, analysis of the Noise2Inverse conditions was limited to single-slice (2D) parallel-beam tomography 25 , treating multi-slice (3D) tomographic datasets as a stack of independent 2D problems. In the synchrotron community, however, 3D tomography is routine and many important research questions can only be answered using more advanced imaging techniques that make use of additional information in the time or spectral domain of the signal 4 – 10 , 26 – 30 .…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

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

Hendriksen

Bührer

Leone

et al. 2021

Sci Rep

Self Cite

View full text Add to dashboard Cite

Synchrotron X-ray tomography enables the examination of the internal structure of materials at submicron spatial resolution and subsecond temporal resolution. Unavoidable experimental constraints can impose dose and time limits on the measurements, introducing noise in the reconstructed images. Convolutional neural networks (CNNs) have emerged as a powerful tool to remove noise from reconstructed images. However, their training typically requires collecting a dataset of paired noisy and high-quality measurements, which is a major obstacle to their use in practice. To circumvent this problem, methods for CNN-based denoising have recently been proposed that require no separate training data beyond the already available noisy reconstructions. Among these, the Noise2Inverse method is specifically designed for tomography and related inverse problems. To date, applications of Noise2Inverse have only taken into account 2D spatial information. In this paper, we expand the application of Noise2Inverse in space, time, and spectrum-like domains. This development enhances applications to static and dynamic micro-tomography as well as X-ray diffraction tomography. Results on real-world datasets establish that Noise2Inverse is capable of accurate denoising and enables a substantial reduction in acquisition time while maintaining image quality.

show abstract

X‐ray Tomography and Tomoscopy on Metals: A Review

et al. 2023

View full text Add to dashboard Cite

X‐ray tomography is a versatile tool in materials research and engineering since it allows for a non‐destructive and three‐dimensional mapping of the constituents of a heterogeneous material as long as they differ in their interactions with X‐rays. Recent developments of the technique have brought down the time needed for the acquisition of a single tomogram by many orders of magnitude compared to what was needed 25 years ago. Nowadays, up to 1000 full tomograms can be recorded in a second, which enables real‐time studies of changes in samples caused by reactions or by applied processing operations. The term tomoscopy has been coined for such sequences of 3D images. We review the application of X‐ray tomography and tomoscopy on metals and describe each step required and the associated challenges. A selection of representative investigations is presented with a focus on time‐resolved phenomena in metals and alloys ranging from mechanical deformation, solidification to metals processing processes such as welding and additive manufacturing. Finally likely future developments are discussed.

show abstract

Noise2Inverse: Self-Supervised Deep Convolutional Denoising for Tomography

Cited by 94 publications

References 48 publications

Clinical Micro-CT Empowered by Interior Tomography, Robotic Scanning, and Deep Learning

Clinical Micro-CT Empowered by Interior Tomography, Robotic Scanning, and Deep Learning

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

X‐ray Tomography and Tomoscopy on Metals: A Review

Contact Info

Product

Resources

About