Limits...
A dictionary learning approach with overlap for the low dose computed tomography reconstruction and its vectorial application to differential phase tomography.

Mirone A, Brun E, Coan P - PLoS ONE (2014)

Bottom Line: The solution is found by applying the iterative proximal gradient descent method with FISTA acceleration.For this case we use an original approach which consists in using vectorial patches, each patch having two components: one per each gradient component.The resulting algorithm, implemented in the European Synchrotron Radiation Facility tomography reconstruction code PyHST, has proven to be efficient and well-adapted to strongly reduce the required dose and the number of projections in medical tomography.

View Article: PubMed Central - PubMed

Affiliation: European Synchrotron Radiation Facility, BP 220, F-38043, Grenoble, France.

ABSTRACT
X-ray based Phase-Contrast Imaging (PCI) techniques have been demonstrated to enhance the visualization of soft tissues in comparison to conventional imaging methods. Nevertheless the delivered dose as reported in the literature of biomedical PCI applications often equals or exceeds the limits prescribed in clinical diagnostics. The optimization of new computed tomography strategies which include the development and implementation of advanced image reconstruction procedures is thus a key aspect. In this scenario, we implemented a dictionary learning method with a new form of convex functional. This functional contains in addition to the usual sparsity inducing and fidelity terms, a new term which forces similarity between overlapping patches in the superimposed regions. The functional depends on two free regularization parameters: a coefficient multiplying the sparsity-inducing L1 norm of the patch basis functions coefficients, and a coefficient multiplying the L2 norm of the differences between patches in the overlapping regions. The solution is found by applying the iterative proximal gradient descent method with FISTA acceleration. The gradient is computed by calculating projection of the solution and its error backprojection at each iterative step. We study the quality of the solution, as a function of the regularization parameters and noise, on synthetic data for which the solution is a-priori known. We apply the method on experimental data in the case of Differential Phase Tomography. For this case we use an original approach which consists in using vectorial patches, each patch having two components: one per each gradient component. The resulting algorithm, implemented in the European Synchrotron Radiation Facility tomography reconstruction code PyHST, has proven to be efficient and well-adapted to strongly reduce the required dose and the number of projections in medical tomography.

Show MeSH
Reconstruction of a computed tomographic slice of the breast.The images on the first and second row are the X and Y phase gradients, respectively. In the left column the results of the reconstruction obtained with the FBP method using the full set of data are reported. In the right column the results of our method using one projection over five are shown. For these reconstructions we set  and .
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4274000&req=5

pone-0114325-g007: Reconstruction of a computed tomographic slice of the breast.The images on the first and second row are the X and Y phase gradients, respectively. In the left column the results of the reconstruction obtained with the FBP method using the full set of data are reported. In the right column the results of our method using one projection over five are shown. For these reconstructions we set and .

Mentions: The result of reconstruction obtained by using filtered back projection algorithm with 1000 projections is shown in fig. 7a. In this image, radiologists could easily identify the skin, fat and glandular tissue. Fig. 7 is the reconstruction of a pixel slice, using only 200 projections over the 1000 available. The upper left square is a zoom in the region marked in subfigure 7. The used projections cover, with constant spacing, a degree range. The right column is the reconstruction with our method for X and Y components, while the left column (subfigure 7c) and d) is reconstructed with the standard FBP using all 1000 available projections. Using our method, we can still generate a high quality image with only one fifth of projections which would otherwise be necessary to generate a high quality reconstruction with the standard FBP method. Visually, the difference between the FBP results obtained with the full data set and our method with a five-fold reduction of the data is barely noticeable. The different borders of structures like skin layers, fatty tissues, and collagen strands are easily identified. The obtained results are very promising and a systematic evaluation for clinical application is under-way. The radiation dose absorbed by the sample during 200 projections is comparable to that of a standard clinical dual view (2D) mammography (3.5 mGy).


A dictionary learning approach with overlap for the low dose computed tomography reconstruction and its vectorial application to differential phase tomography.

Mirone A, Brun E, Coan P - PLoS ONE (2014)

Reconstruction of a computed tomographic slice of the breast.The images on the first and second row are the X and Y phase gradients, respectively. In the left column the results of the reconstruction obtained with the FBP method using the full set of data are reported. In the right column the results of our method using one projection over five are shown. For these reconstructions we set  and .
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4274000&req=5

pone-0114325-g007: Reconstruction of a computed tomographic slice of the breast.The images on the first and second row are the X and Y phase gradients, respectively. In the left column the results of the reconstruction obtained with the FBP method using the full set of data are reported. In the right column the results of our method using one projection over five are shown. For these reconstructions we set and .
Mentions: The result of reconstruction obtained by using filtered back projection algorithm with 1000 projections is shown in fig. 7a. In this image, radiologists could easily identify the skin, fat and glandular tissue. Fig. 7 is the reconstruction of a pixel slice, using only 200 projections over the 1000 available. The upper left square is a zoom in the region marked in subfigure 7. The used projections cover, with constant spacing, a degree range. The right column is the reconstruction with our method for X and Y components, while the left column (subfigure 7c) and d) is reconstructed with the standard FBP using all 1000 available projections. Using our method, we can still generate a high quality image with only one fifth of projections which would otherwise be necessary to generate a high quality reconstruction with the standard FBP method. Visually, the difference between the FBP results obtained with the full data set and our method with a five-fold reduction of the data is barely noticeable. The different borders of structures like skin layers, fatty tissues, and collagen strands are easily identified. The obtained results are very promising and a systematic evaluation for clinical application is under-way. The radiation dose absorbed by the sample during 200 projections is comparable to that of a standard clinical dual view (2D) mammography (3.5 mGy).

Bottom Line: The solution is found by applying the iterative proximal gradient descent method with FISTA acceleration.For this case we use an original approach which consists in using vectorial patches, each patch having two components: one per each gradient component.The resulting algorithm, implemented in the European Synchrotron Radiation Facility tomography reconstruction code PyHST, has proven to be efficient and well-adapted to strongly reduce the required dose and the number of projections in medical tomography.

View Article: PubMed Central - PubMed

Affiliation: European Synchrotron Radiation Facility, BP 220, F-38043, Grenoble, France.

ABSTRACT
X-ray based Phase-Contrast Imaging (PCI) techniques have been demonstrated to enhance the visualization of soft tissues in comparison to conventional imaging methods. Nevertheless the delivered dose as reported in the literature of biomedical PCI applications often equals or exceeds the limits prescribed in clinical diagnostics. The optimization of new computed tomography strategies which include the development and implementation of advanced image reconstruction procedures is thus a key aspect. In this scenario, we implemented a dictionary learning method with a new form of convex functional. This functional contains in addition to the usual sparsity inducing and fidelity terms, a new term which forces similarity between overlapping patches in the superimposed regions. The functional depends on two free regularization parameters: a coefficient multiplying the sparsity-inducing L1 norm of the patch basis functions coefficients, and a coefficient multiplying the L2 norm of the differences between patches in the overlapping regions. The solution is found by applying the iterative proximal gradient descent method with FISTA acceleration. The gradient is computed by calculating projection of the solution and its error backprojection at each iterative step. We study the quality of the solution, as a function of the regularization parameters and noise, on synthetic data for which the solution is a-priori known. We apply the method on experimental data in the case of Differential Phase Tomography. For this case we use an original approach which consists in using vectorial patches, each patch having two components: one per each gradient component. The resulting algorithm, implemented in the European Synchrotron Radiation Facility tomography reconstruction code PyHST, has proven to be efficient and well-adapted to strongly reduce the required dose and the number of projections in medical tomography.

Show MeSH