Marcel Van Straten PhD, Presenter: Research collaboration, Siemens AG

Gabriel P. Krestin MD, PhD, Abstract Co-Author: Consultant, General Electric Company Research Grant, General Electric Company Research Grant, Bayer AG Research Grant, Siemens AG Speakers Bureau Siemens AG

To validate an algorithm for reduced dose simulations and to study the clinical feasibility of dose reduction by iterative reconstruction (IR) techniques.

The algorithm converts first each attenuation value to a number N_a of detected photons based on calibration measurements taking tube current modulation, bowtie filtration, and beam hardening into account. Next, N_a is reduced to N_b = alpha * Poisson(beta * N_a). Based on the rules for conditional expectation and variance the factors alpha and beta are derived from a model describing quantum noise, electronic noise, and noise due to the polychromatic x-ray spectrum. Finally, the simulated attenuation values are passed back to the scanner to reconstruct low dose images. Reading and writing of sinogram data was performed with vendor-supplied software. For the validation, an anthropomorphic thorax and abdomen phantom were scanned on a dual source scanner (Definition Flash, Siemens) at various exposure values (30-240 mAs). Dose reduction potential of the IR technique (SAFIRE) was assessed with patient scans of various body parts, including thorax and abdomen. Contrast dependent sharpness and lesion detectability were evaluated as a function of simulated dose reduction and IR technique strength setting.

No significant difference was found for the standard deviation of the image noise between measured and simulated low dose phantom scans (p<0.01). The shape of the noise power spectrum was not affected by the algorithm. In patients, the IR at maximum strength was able to compensate for the image noise increase due to a simulated dose reduction of a factor 3-4, depending on the reconstruction kernel used. This was accompanied by a change in image impression due to a change of the noise power spectrum. Subtraction images revealed slightly improved high contrast sharpness in simulated low dose scans with IR compared to measured normal dose scans with filtered backprojection.

In phantoms, simulations were in good agreement with measurements at reduced dose. In patients, influence of dose reduction and IR on image impression, noise texture and low contrast detectability could be assessed.

Scanning patients both at normal and low dose is unethical and unpractical. Phantoms, however, lack realistic tissues. Low dose simulations offer an alternative to evaluate iterative reconstructions.

Van Straten, M, Krestin, G, Assessment of the Dose Reduction Potential of Iterative Reconstruction Techniques with a Validated Algorithm for Sinogram-based Simulation of Low Dose Scans in Patients.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14009077.html