Fast and flexible X-ray tomography using the ASTRA toolbox

Aarle, Wim van; Palenstijn, Willem Jan; Cant, Jeroen; Janssens, Ewald; Bleichrodt, Folkert; Dabravolski, Andrei; Beenhouwer, Jan De; Batenburg, Kees Joost; Sijbers, Jan

doi:10.1364/oe.24.025129

Cited by 838 publications

(573 citation statements)

References 23 publications

Supporting

Mentioning

519

Contrasting

Unclassified

Order By: Relevance

“…The implementation utilizes abstract ODL structures for representing functional analytic notions and is therefore generic, yet easily adaptable to other inverse problems. We used the ODL operator RayTransform in order to evaluate the ray transform and its adjoint using the GPU accelerated 'astra gpu' backend [37]. The pseudo-inverse T † was given by the filtered back-projection algorithm implemented in ODL as fbp op with no additional smoothing filter.…”

Section: Methodsmentioning

confidence: 99%

Solving ill-posed inverse problems using iterative deep neural networks

2017

View full text Add to dashboard Cite

We propose a partially learned approach for the solution of ill posed inverse problems with not necessarily linear forward operators. The method builds on ideas from classical regularization theory and recent advances in deep learning to perform learning while making use of prior information about the inverse problem encoded in the forward operator, noise model and a regularizing functional. The method results in a gradient-like iterative scheme, where the "gradient" component is learned using a convolutional network that includes the gradients of the data discrepancy and regularizer as input in each iteration.We present results of such a partially learned gradient scheme on a non-linear tomographic inversion problem with simulated data from both the Sheep-Logan phantom as well as a head CT. The outcome is compared against filtered backprojection and total variation reconstruction and the proposed method provides a 5.4 dB PSNR improvement over the total variation reconstruction while being significantly faster, giving reconstructions of 512 × 512 pixel images in about 0.4 s using a single graphics processing unit (GPU).

show abstract

Section: Methodsmentioning

confidence: 99%

Solving ill-posed inverse problems using iterative deep neural networks

2017

View full text Add to dashboard Cite

show abstract

“…SART is known for performing well when the projection data is limited, and DART supplements this with an assumption on only a few attenuation levels. We used the implementations provided in the ASTRA toolbox [34,35]. For this experiment we converted our test object into a 500 × 500 pixels binary image.…”

Section: Resultsmentioning

confidence: 99%

Computing segmentations directly from x-ray projection data via parametric deformable curves

Dahl¹,

Dahl²,

Hansen³

2017

Meas. Sci. Technol.

View full text Add to dashboard Cite

We describe an efficient algorithm that computes a segmented reconstruction directly from x-ray projection data. Our algorithm uses a parametric curve to define the segmentation. Unlike similar approaches which are based on level-sets, our method avoids a pixel or voxel grid; hence the number of unknowns is reduced to the set of points that define the curve, and attenuation coefficients of the segments. Our current implementation uses a simple closed curve and is capable of separating one object from the background. However, our basic algorithm can be applied to an arbitrary topology and multiple objects corresponding to different attenuation coefficients in the reconstruction. Through systematic tests we demonstrate a high robustness to the noise, and an excellent performance under a small number of projections.

show abstract

“…In which case, the application of iterative algorithms like the Simultaneous Iterative Reconstruction Technique (SIRT) [20], the Simultaneous Algebraic Reconstruction Technique (SART) [21], or other more complex techniques [22] for tomography reconstruction calculations allows qualitative data to be obtained. The ASTRA Toolbox [23] is a MATLAB and Python platform providing high-performance GPU primitives for 2D and 3D A set of neutron radiography images was collected by a CCD-based detector system with a maximum field of view of 20 cm × 20 cm. A dedicated detector module with two mirrors to suppress parasitic neutron scattering was used ( Figure 2).…”

Section: Methodsmentioning

confidence: 99%

“…The reduction of calculation time is possible by using specialized professional graphics cards with CUDA technology [24] support. However, in order to use the ASTRA toolbox, the user requires the skills necessary to parse and process data in the MATLAB or Python layer [23,25]. Moreover, the algorithms bundled in the ASTRA Toolbox are limited to reconstruction methods, and do not include typical image pre-processing, like flat field correction or rotation center calculation algorithms.…”

Section: Methodsmentioning

confidence: 99%

Studies of Ancient Russian Cultural Objects Using the Neutron Tomography Method

et al. 2018

View full text Add to dashboard Cite

Neutron radiography and tomography is a non-destructive method that provides detailed information about the internal structure of cultural heritage objects. The differences in the neutron attenuation coefficients of constituent elements of the studied objects, as well as the application of modern mathematical algorithms to carry out three-dimensional imaging data analysis, allow one to obtain unique information about the spatial distribution of different phases, the presence of internal defects, or the degree of structural degradation inside valuable cultural objects. The results of the neutron studies of several archaeological objects related to different epochs of the Russian history are reported in order to demonstrate the opportunities provided by the neutron tomography method. The obtained 3D structural volume data, as well as the results of the corresponding data analysis, are presented.

show abstract

Fast and flexible X-ray tomography using the ASTRA toolbox

Cited by 838 publications

References 23 publications

Solving ill-posed inverse problems using iterative deep neural networks

Solving ill-posed inverse problems using iterative deep neural networks

Computing segmentations directly from x-ray projection data via parametric deformable curves

Studies of Ancient Russian Cultural Objects Using the Neutron Tomography Method

Contact Info

Product

Resources

About