Herman van Langen PhD, Presenter: Nothing to Disclose

KlaasJan Renema PHD, Abstract Co-Author: Nothing to Disclose

In digital projection radiology the image quality can be determined using a contrast detail phantom. Traditionally, human observers perform analysis of images made with this phantom by indicating all visible spots. It is time-consuming and the method has shown to be cumbersome due to considerable inter and intra individual variations. Dedicated software is now commercial available to determine which spots on the phantom image are detectable and which aren’t. Its utility for assessing the quality of an image in an observer independent way will be tested.

The contrast detail phantom CDRAD has been used. It consists of a Perspex tablet with 15 rows and 15 columns. In each cell cylindrical holes are drilled which vary in diameter (0,3 to 8 mm) for detail and depth (0,3 to 8 mm) for contrast purposes. Using a Siemens FD-X flat panel detector, 9 sets of 5 images were made. The impact of the addition of Perspex material, a grid and the automatic exposure control (AEC) on the image quality is tested. The inverse image quality factor (IQi) is taken as the outcome of the CDRAD Analyzer software (Artinis). IQi is computed from all detected spots and it increases with the number of correctly identified spots.

For all images the IQi ranged from 1,8 to 8,5. For the 9 sets, the average coefficient of variation for IQi was 7%. With manual exposure settings •4 cm Perspex decreases the IQi from 8,2±0,2 to 7,1±0,4 (46 kV, no grid) •11 cm Perspex decreases the IQi from 8,0±0,2 to 3,4±0,2 (50 kV, with grid)•11 cm Perspex decreases the IQi from 7,8±0,0 to 3,2±0,2(50 kV, no grid) •Use of a grid decreases the IQi from 7,1±0,4 to 6,3±0,6 (46 kV, 4 cm Perspex. With AEC, the use of a grid increases IQi from 2,5±0,5 to 5,8±0,7 with a concomitant fourfold increase in mAs. These findings are similar to analysis performed by human observers.

The CDRAD Analyzer is a promising tool to determine image quality in an observer independent way. It can be useful to determine the dose and image quality relation. It may have several clinical applications but more validation studies are needed.

van Langen, H, Renema, K, In Digital Projection Radiology Image Quality Can Be Determined in an Observer Independent Way.  Radiological Society of North America 2005 Scientific Assembly and Annual Meeting, November 27 - December 2, 2005 ,Chicago IL. http://archive.rsna.org/2005/4404735.html