Objective: the objective of this ROC-study was to evaluate the diagnostic efficacy of images acquired with a grid in digital selenium radiography compared to that on images obtained with the integrated air gap only.
Materials and methods: seven types of simulated lesions were superimposed onto an anthropomorphic chest phantom. Selenium radiography images were obtained either with or without an additional antiscatter grid. For images acquired with a grid either a similar or increased exposure level was used. Both normal and obese patients were simulated.
Results: When a grid was used with an equivalent detector dose and a higher exposure, diagnostic performance was significantly improved as compared to images obtained with only the air gap. ROC curve areas for mediastinal nodules and catheters were substantially higher for images acquired with a grid and the same exposure level compared to images obtained without a grid. However, detection of linear, net-shaped and reticulonodular structures in peripheral lung regions was significantly worse when a grid was used with an equivalent exposure level. Concerning the interpretation of images obtained from the normal and obese phantom models, no substantial differences were observed.
Conclusion: a marked improvement in diagnostic performance could be achieved by means of the use of an additional antiscatter grid and an equivalent detector dose. However, when the same exposure was used, images acquired with the grid allowed a better detection of mediastinal structures although a worse performance was evident in radiolucent lung regions. Therefore, the routine use of a grid without increased exposure is not recommended.