Optimal kVp in chest computed radiography using visual grading scores: a comparison between visual grading characteristics and ordinal regression analysis

Abstract

The purposes of this work were to determine the optimal peak voltage for chest computed radiography (CR) using visual grading scores and to compare visual grading characteristics (VGC) and ordinal regression in visual grading analysis. An Afga CR system was used to acquire images of an anthropomorphic chest phantom. Both entrance surface dose and detector surface dose were measured using the Piranha 657 dosimeter. The images were acquired under various voltages from 80 to 120 kVp and exposures from 0.5 to 12.5 mAs. The image qualities were evaluated by 5 experienced radiologists/radiographers based on modified European imaging criteria using 1-5 visual grading scale. The VGC, ordinal regression as well as the conventional visual grading analysis (VGA) were employed for the image quality analysis. Both VGC and ordinal regression yielded the same results with both 100 kVp and 120 kVp producing the best image quality. The image quality of the 120 kVp was slightly higher than that of the 100 kVp but its dose was also higher than that of the 100kVp. On balancing image quality with dose, the 100 kVp should be the optimal kVp for the chest imaging using the Afga CR system. The ordinal regression is a powerful tool in the analysis of image quality using visual grading scores and the VGC can be handled by the ordinal regression.