Detection of simulated periodontal bone gain by digital subtraction radiography with tuned-aperture computed tomography. The effect of angular disparity.

Abstract

To investigate the influence of angular disparity on observer detection of simulated bone gain in digital subtraction radiography using tuned-aperture computed tomography (TACT). Simulated periodontal defects were created in interproximal and buccal or lingual (tooth-obscured) areas of the premolar and molar regions of a dry human skull. Radiographs were obtained before and after known weights of amorphous bone were added to the defects to simulate bone gain. The skull was positioned in a multidirectional tomographic unit to achieve reproducibility. A series of nine basis images were acquired with a CMOS intra-oral receptor and repeated using angular disparities of 10 degrees, 20 degrees, and 30 degrees. Stacks of TACT slices generated from the basis images were paired for image-registration, histogram-equalization and subtraction using TACT Workbench. Eight calibrated observers randomly assessed the presence or absence of bone gain using a 5-point confidence scale. ROC curves were generated and A(z) values were analysed using ANOVA. There were significant differences in the performance of the observers (P=0.034), defect location (P=0.005), amount of bone gain (P<0.001), angular disparity (P=0.003) and angular disparity x defect location interaction (P=0.019). Mean A(z) values in detecting bone gain were 0.90, 0.85, 0.79 for angular disparities of 10 degrees, 20 degrees, and 30 degrees respectively. Smaller angular disparity provided better detection of bone gain with TACT-subtraction using nine basis-projections. This effect of angular disparity was especially evident with tooth-obscured defects.