Blur Reduction of Conventional Film-Based Tomograms For Pre-Surgical Evaluation of Potential Mandibular Implant Sites

Abstract

The usefulness of motion-based cross-sectional tomography to evaluate osseous support and adjacent anatomical structures for dental implant placement is limited by the inherent blurring in these images. The goal of this study was to develop a method to remove blurring while permitting accurate dimensional analysis of the potential implant site. Defined regions (anterior, cuspid, premolar, molar) on two preserved human mandibles were imaged using cross-sectional linear tomography. Algorithms were developed as a personal computer application to remove the blur and to aid in identification of the cortical plate borders and the mandibular canal. The data set of eight tomograms was digitized and the blur reduced with the developed algorithm. An operator measured the height and width of the mandible on each original tomogram and each deblurred tomogram in triplicate. Method error was calculated as the difference between direct caliper measurements of the respective skull regions and image measurements of height and width for both the original digitized tomograms and the deblurred tomograms. Method error using the original images (height: -2.72 +/- 2.15 mm; width: -0.58 +/- 1.36 mm) compared to the deblurred tomograms (height: -0.58 +/- 1.16 mm; width: 0.37 +/- 0.59 mm) was significantly greater for both height (t-test level of significance, P = 0.0047) and width (t-test level of significance, P = 0.0001). These findings suggest that the method developed may greatly improve the ability of clinicians to accurately assess the implant site using cross-sectional film-based linear tomograms.