Simone Eriksson MMEDSC, Abstract Co-Author: Nothing to Disclose

Fredrik Stålhammar MD, Abstract Co-Author: Nothing to Disclose

Lars-Martin Wiklund MD, PhD, Abstract Co-Author: Nothing to Disclose

Liz Ivarsson MD, Abstract Co-Author: Nothing to Disclose

Kerstin Ledenius MS, Abstract Co-Author: Nothing to Disclose

Anne Thilander-Klang PhD, Presenter: Nothing to Disclose

To evaluate the impact of Adaptive Statistical iterative Reconstruction (ASiR) on image quality in pediatric head CT exams.

Prior to the study 30% of ASiR (and 70% of filtered back projection (FBP)) was clinically used, based on experience from adult head CT exams. Original digital scanning data from 19 patients (mean age 11 years; range 5 to 17 years) undergoing routine head CT on a 64 slice MDCT scanner (GE Discovery CT750 HD) were retrospectively reconstructed using 0, 20, 30, 40, 50, 60 and 100% of ASiR. Three experienced pediatric radiologists assessed the 5 mm thick axial images in stacks in a blinded randomized visualization study. The reproduction of six anatomical structures in the brain and the overall diagnostic acceptability were assessed using a four point verbal rating scale. Data were analyzed using a method for paired ordinal data that identifies and measures systematic disagreement separately from individual variations in paired assessments. Noise, determined as the standard deviation of the CT-value in Hounsfield Units (HU), was measured for each image stack.

Frequency distribution of all the assessments for all observers showed that 50% of ASiR generated the highest ranking scores. Comparison of the overall diagnostic acceptability between 30% and 50% of ASiR resulted in a Relative Position (RP) value of -0.21 (95% CI -0.34 to -0.07), which means that it was rated significantly higher for 50% of ASiR. The visibility of the basal ganglia improved with increased ASiR-percentage while the delineation of the cerebrospinal fluid around the brain decreased when ASiR of 60% or more was used. Intraobserver agreement regarding all evaluation criteria was 45%, 81% and 85% for observer 1, 2 and 3, respectively. Interobserver agreement regarding the overall diagnostic acceptability for observer 1 vs. 2 was 54%, 1 vs. 3 was 56 % and 2 vs. 3 was 65%. The image noise level for 50% of ASiR was 2.7 ± 0.2 HU compared to the level of 30% of ASiR, which was 3.1 ± 0.3 HU.  

Using 50% of ASiR improves image quality of pediatric head CT exams at the same dose level used at 30% of ASiR (original images). This indicates that further dose reduction is possible.

Using the optimal combination of ASiR and filtered back projection (FBP) will possibly increase the diagnostic quality in pediatric head CT exams and enable dose reduction at preserved image quality.

http://media.rsna.org/media/abstract/2010/9002504/zpbb781_9002504_uplok.pdf

Eriksson, S, Stålhammar, F, Wiklund, L, Ivarsson, L, Ledenius, K, Thilander-Klang, A, Impact of Different Levels of Adaptive Statistical Iterative Reconstruction on Image Quality in Pediatric Head CT Exams.  Radiological Society of North America 2010 Scientific Assembly and Annual Meeting, November 28 - December 3, 2010 ,Chicago IL. http://archive.rsna.org/2010/9002504.html