Encrypted login | home

Program Information

A Simple Method for Simulating Reduced-Dose Images for Evaluation of Clinical CT Protocols


N Bevins

N Bevins1*, T Szczykutowicz2, MP Supanich1, (1) Henry Ford Health System, Detroit, MI, (2) University Wisconsin-Madison, Madison, WI

TU-C-103-6 Tuesday 10:30AM - 12:30PM Room: 103

Purpose: To develop a simple method using readily-available software and equipment to simulate clinically-realistic dose reduction levels for CT images from regular-dose clinical images. The resulting images will be used to demonstrate the effects of dose reduction on image quality to clinicians and aid in guiding protocol revision.

Methods: A reconstruction of an anthropomorphic phantom scanned under a clinical protocol was imported into MATLAB. The image was converted to attenuation coefficients before forward-projecting using the built-in radon function. Noise was added to the projections using a Poisson statistics model. The original radon projections were then subtracted from the simulated projections to produce noise-only projections, which could then be reconstructed using the iradon function with a custom filter. The filter was calculated by first measuring the radial average of the two-dimensional noise power spectrum (NPS) in a noise-only image generated from subtracted back-to-back scans of a water cylinder. The filter used by the scanner is proportional to the square-root of the radially-averaged NPS after normalizing it to radial frequency. The final noise-only reconstruction is then added to the original clinical image to simulate a reduced-dose scan.

Results: The noise-only reconstructions show accurate structure and texture for both the standard and bone reconstructions of the head phantom. The final simulated images (with the noise reconstruction added to the original reconstruction) closely match the appearance of the actual reduced-dose images.

Conclusion: Previous methods for noise simulation that rely on raw data access and advanced system knowledge allow for very low dose levels to be accurately simulated. However, these ultra-low dose levels are often too low to be of clinical significance. This work provides a simple and accurate tool for dose reduction simulations that are more clinically realistic. In addition, this method only requires measurements that can be acquired using standard procedures and phantoms.

Contact Email: