Evaluation of CDRAD and TO20 test objects and associated software in digital radiography

Adequate reproduction of low contrast image detail is essential for accurate diagnosis in diagnostic radiology. Artinis CDRAD and Leeds Test Object TO20 are commercially available test objects that can be used to test this aspect of imaging performance in digital radiography. Automated analysis software is available for both test objects (CDRAD Analyser and AutoPIA (TO20)). This study evaluated and compared both test objects and software, including their sensitivity to changes in exposure parameters and image processing. Images of the test objects were acquired in scatter and scatter-free environments and analysed using a range of metrics derived from automated analysis, visual assessment and manual contrast-to-noise ratio measurements. The CNR (TO20) and correctly identified holes (%) (CDRAD) were found to be the most sensitive to changes in exposure conditions. The total number of detected discs (TO20) was the least sensitive. None of the image quality metrics was sensitive to a change in image processing. Other methods, such as anthropomorphic phantoms, would therefore be required to evaluate changes in clinical image processing. The results of this investigation help to inform best use of these test objects in routine quality control and image quality benchmarking for optimisation.