Unmet Need
Computed tomography (CT) use has dramatically increased over the past 20 years such that around 31% of primary care patients undergo CT scans every year. All CT scans employ “back-projection” as the basic computational method to relate 2D detector signal values into a 3D re-construction. However, image reconstruction sometimes leads to noise and artifacts in the im-age and alternative methods for reconstruction have caused major increase in computational burden. Thus, there is a need for a 3D image reconstruction method for CT images that de-creases noise and artifacts within the image, but without drastically increasing computational burden.
Technology Overview
Johns Hopkins researchers have proposed an invention called stochastic backprojector. This method modifies an existing backprojection operator to alleviate the impact of sampling artifacts with very minimal computational burden. The method consists of random perturbations of the ray position in each voxel. The perturbations are computed for every ray traced in the recon-struction problem. For the projection angles, new ray intersect positions are computed that can then be used to tune the backprojection to achieve the desire balance between sampling pertur-bation and fidelity.
Stage of Development
The method is currently undergoing testing.