Objective evaluation method using multiple image analyses for panoramic radiography improvement

Introduction: In the standardization of panoramic radiography quality, the education and training of beginners on panoramic radiographic imaging are important. We evaluated the relationship between positioning error factors and multiple image analysis results for reproducible panoramic radiography.
Material and methods: Using a panoramic radiography system and a dental phantom, reference images were acquired on the Frankfurt plane along the horizontal direction, midsagittal plane along the left-right direction, and for the canine on the forward-backward plane. Images with positioning errors were acquired with 1-5 mm shifts along the forward-backward direction and 2-10 degrees rotations along the horizontal (chin tipped high/low) and vertical (left-right side tilt) directions on the Frankfurt plane. The cross-correlation coefficient and angle difference of the occlusion congruent plane profile between the reference and positioning error images, peak signal-to-noise ratio (PSNR), and deformation vector value by deformable image registration were compared and evaluated.
Results: The cross-correlation coefficients of the occlusal plane profiles showed the greatest change in the chin tipped high images and became negatively correlated from 6 degrees image rotation (r = -0.29). The angle difference tended to shift substantially with increasing positioning error, with an angle difference of 8.9 degrees for the 10 degrees chin tipped low image. The PSNR was above 30 dB only for images with a 1-mm backward shift. The positioning error owing to the vertical rotation was the largest for the deformation vector value.
Conclusions: Multiple image analyses allow to determine factors contributing to positioning errors in panoramic radiography and may enable error correction. This study based on phantom imaging can support the education of beginners regarding panoramic radiography.