Digital image analysis of cadaver mandibular trabecular bone patterns

The purpose of this study was to investigate the source of radiographic trabecular patterns by removing trabecular bone in four sequential steps from six cadaver mandible sections, radiographing the sections after each removal, and using four digital-image analysis methods to quantify any resulting changes to the radiographs.

Mandible sections were cut sagittally into halves. Trabecular bone was removed from each section in four stages. Following each stage, standardized radiographs were taken, using direct digital equipment. Trabecular bone in the resulting digital images was measured with four methods. Mean gray level values (method 1) and cumulative percent histograms (method 2) were calculated from the raw data. Morphological image processing was used to skeletonize the trabecular structure, which was quantified by counting the number of trabecular ends and segments in the skeletonized images (method 3) and performing fractal analyses of the skeletonized images (method 4). Repeated measures analysis of variance (ANOVA) was used to test for changes in measurements attributable to bone removal.

Repeated measures ANOVA indicated that the use of gray levels, cumulative percent histograms, and morphologic operators resulted in highly significant changes in measurements following bone removal (P < 0.01). Ends and segments demonstrated similar performance, with changes highly significant over time (P < 0.01). Fractal analysis also resulted in highly significant changes over time (P < 0.01).

The analyses performed in this study demonstrated consistent image differences following the four steps of bone removal. These differences appeared whether light, cancellous bone or heavier endosteal bone was removed. These findings indicate that trabecular and endosteal bone combine to form the structure that most dentists identify as trabeculae on intraoral radiographs.