Nicholas Keat MSc, BS, Presenter: Nothing to Disclose

Eugenia Kulama MSc, Abstract Co-Author: Nothing to Disclose

Olivia Egan MS, Abstract Co-Author: Nothing to Disclose

Two techniques for iterative reconstruction of CT scan data were assessed by studying changes to noise, noise power spectrum (NPS) and modulation transfer function (MTF) under a variety of conditions. As the effect of iterative techniques can vary spatially, each image quality metric was assessed from the same images over a region less than 50 mm in size.

Two Siemens Definition CT scanners equipped with image based (IRIS) and raw data based (SAFIRE) iterative reconstruction capabilities were assessed. A 100 mm diameter solid cylinder containing a 100 µm nickel wire was placed in water filled head and anthropomorphic body phantoms. The phantoms were scanned using standard spiral head and body protocols, and at reduced x-ray tube current. 20 images were reconstructed from each scan with standard and sharp clinical kernels, with and without iterative reconstruction at fields of view from 50 to 300 mm (head) or 50 to 500 mm (body). Image noise was assessed by the standard deviation of an annular ROI centred on the wire, with internal and external diameters of 20 and 50 mm. The mean of the 20 images was subtracted from each image to remove the wire and low frequency non-uniformity. The NPS was then obtained from a square region of interest centred on the wire. The MTF was generated from the point spread function of the wire, and MTF50 and MTF10 values were calculated.

Noise in the iteratively reconstructed images was lower than the standard reconstruction. For body scanning at standard clinical settings, noise was 30% lower with IRIS, and between 12% and 48% lower with SAFIRE depending upon the chosen 'strength' of the iterative algorithm. For head scanning, the figures were 9% and 7% to 37% lower for IRIS and SAFIRE respectively. For both algorithms, the decrease in noise was dependent upon the field of view and was greater for sharp kernels but did not depend upon the tube current. Changes in noise were reflected by changes to the NPS. The MTF was only minimally affected by the use of iterative reconstruction in all cases.

Two iterative reconstruction techniques were shown to offer significant reductions in image noise, without local loss of spatial resolution.

CT noise reduction can be achieved using iterative reconstruction without the increase in patient radiation dose that would be required to achieve the same result by altering scan technique.

Keat, N, Kulama, E, Egan, O, The Effect of CT Sinogram and Image-based Iterative Reconstructions upon Image Noise, Noise Power Spectrum, and Spatial Resolution.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11008928.html