Sparsity constraints and dedicated acquisition protocols for improved Digital Subtraction Rotational Angiography

Hélène Langet (1, 2) , C. Riddell (3) , Y. Trousset (3) , Elisabeth Lahalle (1) , Arthur Tenenhaus (1) , Gilles Fleury (1) , Nikolaos Paragios (2, 4)

1 E3S - Supélec Sciences des Systèmes
2 MAS - Mathématiques Appliquées aux Systèmes - EA 4037
3 GE Healthcare - General Electric Medical Systems [Buc]
4 GALEN - Organ Modeling through Extraction, Representation and Understanding of Medical Image Content

Digital Subtraction Rotational Angiography (DSRA) allows reconstruction of three-dimensional vascular structures from two spins: the contrast is acquired after injecting vessels with a contrast medium, whereas the mask is acquired in the absence of injection. The vessels are then detected by subtraction of the mask from the contrast. Standard DSRA protocol samples the same set of equiangular-spaced positions for both spins. Due to technical limitations of C-arm systems, streak artifacts degrade the quality of all three reconstructed volumes. Recent developments of compressed sensing have demonstrated that it is possible to recover a signal that is sparse in some basis under limited sampling conditions. In this paper, we propose to improve the reconstruction quality of non-sparse volumes when there exists a sparse combination of these volumes. To this purpose, we develop an extension of iterative filtered backprojection that jointly reconstructs the mask and contrast volumes via '1-minimization of sparse priors. A dedicated protocol based upon interleaving both spins is shown to further benefit from the sparsity assumptions, while using the same total number of measurements. Our approach is evaluated in parallel geometry on simulated phantom data.

Domaines: Imagerie médicale

Hélène Langet, C. Riddell, Y. Trousset, Elisabeth Lahalle, Arthur Tenenhaus, et al.. Sparsity constraints and dedicated acquisition protocols for improved Digital Subtraction Rotational Angiography. 11th International Meeting on Fully Three - Dimensional Image Reconstruction in Radiology and Nuclear Medicine (Fully 3D'11), Jul 2011, Postdam, Germany. pp.427-430. ⟨hal-00639448⟩