Clinical application of a ct dose reduction simulator

Abstract

Purpose To validate an exposure reduction algorithm for head CT examinations in clinical practice. Material and methods CT images from Head Adult protocol in a Siemens CT Somatom Definition AS+ was used for this study. Dose reduction was simulated through the addition of noise to each pixel value of the original image. A homogeneous phantom was used to measure the autocorrelation function, which is convolved with a white noise matrix to get the spatially correlated statistical noise to be added. A Matlab code was written to perform this task. Firstly, the algorithm was validated using two phantoms, i.e. a homogeneous and an anthropomorphic heterogeneous phantom, CT-scanned with different exposure reductions, up to 75%, from the original reference mAs. Noise values comparison between CT-simulated and CT-real images was done. Secondly, from our PACS 10 patients previously CT-scanned, with subtle pathologies and normal brain, were selected. CT-simulated images, up to an exposure reduction of 50%, were reviewed by two experienced radiologists, not aware of the applied exposure reduction. Results The algorithm reproduces well the image noise variation with the reduction of the mAs. Differences were less than 2% in both phantoms in the exposure range analysed. CT-simulated images with exposure reductions equal or larger than 30% were clearly identified by both radiologists. Mostly, a 20% reduction was deemed appropriate for clinical diagnosis. Conclusion Dose reduction simulation software might be a powerful tool for patient dose optimization.