Toshihiro Ishihara RT, Presenter: Nothing to Disclose

Shinosuke Matsumoto, Abstract Co-Author: Nothing to Disclose

Akira Taniguchi RT, Abstract Co-Author: Nothing to Disclose

Kouzou Hanai PhD, Abstract Co-Author: Nothing to Disclose

The spatial resolution of filtered back-projection (FBP) images and reconstructed images obtained using iterative approximation was assessed by comparing a common wire method with a variety of other methods, including the chart method, the low-contrast detectability (LCD) method using materials with various CT values in consideration of noise components, and the edge method.

Adaptive Iterative Dose Reduction (AIDR) incorporated in Aquilion ONE (Toshiba Medical Systems, Otawara, Japan) was employed as the iterative approximation reconstruction method. The modulation transfer function (MTF) with and without AIDR was calculated and compared using the different spatial resolution assessment methods: the wire method (CT-200B), the chart method (Catphan@600), the low-contrast detectability method (MHT 15% 10 mm), and the edge method (custom made). For all assessment methods, three patterns of image SD (10, 30, 50) were prepared. In addition, for the edge method, the CT value of the material for sampling the edges was set to four levels (-15, 50, 100, 300 HU).

Wire method: MTF was the same with AIDR ON and AIDR OFF for all patterns. Chart method: MTF with AIDR ON was lower at SD30 and SD50, but as noise increased, errors occurred in automatic analysis. LCD method: MTF was the same for all patterns regardless of AIDR. Edge method: MTF showed no difference between the two at CT values of 100 and 300 HU. At a CT value of 50 HU, compared with the MTF without AIDR, the MTF with AIDR was lower (SD30, 50% MTF: 2.8%, SD50, 50% MTF: 5.7%). It was also lower at a CT value of -15 HU (SD30, 50% MTF: 23.8%, SD50, 50% MTF: 24.2%).

The wire method is not suitable for assessing spatial resolution of noise-reduced images reconstructed using iterative approximation. In this study, the edge phantom with different noise levels at different CT values was found to be useful for assessing the spatial resolution of noise-reduction images, and it was confirmed that the spatial resolution was reduced with AIDR ON.

Iterative approximation reduces noise, adding processing that maintains spatial resolution, which is unchanged for materials with large differences in CT values, but lower with small differences.

Ishihara, T, Matsumoto, S, Taniguchi, A, Hanai, K, Comparison of Methods for Evaluating Spatial Resolution of Iterative Approximation Reconstructed Images.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11000066.html