Rostyslav Boutchko PhD, Presenter: Nothing to Disclose

In MR angiography technique called VIPR, clinical quality images are obtained from radial projection data significantly undersampled with respect to the Nyquist criterion. To test if the undersampled approach can be used in cone-beam computed tomography (CBCT), we construct a reconstruction algorithm for uniform distribution of X-ray foci on a surface of a sphere.

Most reconstruction algorithms in CBCT are derived for sources distributed along a piecewise continuous curve. This trajectory choice imposes certain limitations on selection of the optimal scanning geometry for undersampling the projection data.To test the possibility of undersampling in CT, we propose to consider sources uniformly distributed on a surface. The exact reconstruction formula is derived for the simple model when every point on a spherical shell surrounding the imaged object is a focus of an X-ray cone-beam. Taking into the account data redundancy, we develop a filtered backprojection (FBP) reconstruction algorithm. To implement the algorithm, we select a number of focal spots uniformly distributed on a sphere. For each focal spot, we acquire the projections on a planar detector grid, filter them and back project them into the ROI with 1/R³ weight.

For 512 x 512 and 1024 x 1024 detectors, we are able to reconstruct simulated projection data for as few as 100-300 separate views. We present the reconstructed images for different phantoms, study the signal fluctuations due to the artifacts caused by undersampling, and demonstrate detail reconstruction from noisy data. We show how our reconstruction approach can be applied to other source distribution geometries.

We have created a new FBP reconstruction algorithm in CBCT for sources distributed on a sphere and demonstrated the possibility of reconstructing tomographic images from very sparsely sampled projection data. The proposed algorithm can be extended to realistic scanning geometries and used to improve acquisition time and image quality in CT angiography, cardiac imaging, and other volumetric CT studies.

Boutchko, R, Simulations of Undersampled Cone-beam CT Using a Spherical Source Distribution.  Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL. http://archive.rsna.org/2004/4412026.html