Fractal analysis of dental periapical radiographs: A revised image processing method.

OBJECTIVE

To assess trabecular bone structure as calculated with fractal analysis by 2 binarization processes: White and Rudolph's original method (WR.o) and a revised version (WR.r). Fractal dimension (FD) values calculated with WR.r (FD.r) and a gray-scale-based method (FD.f) were also compared. FD, histogram parameters, and lacunarity were compared by dentate status, jaw location, and sex.

STUDY

DESIGN

Regions of interest from digital periapical radiographs were defined below the teeth roots and in the edentulous sites of 37 patients. Histograms were assessed for pixel values. Binarization was performed with WR.o and then with WR.r, in which the outliers were removed. FD was assessed using WR.r (FD.r) and (FD.f). Histograms were assessed to obtain pixel values. Lacunarity was calculated.

RESULTS

WR.r revealed fewer trabeculae, branches, and junctions than WR.o (P < .0001). The majority of the mean differences between FD.r and FD.f were within the 95% CI. Dentate areas had greater mean gray levels than partially edentulous areas (P = .0027). FD.f was higher in the mandible (P = .01), but gray-level SD (P < .0001) and lacunarity (P = .02) were greater in the maxilla. FD.f and lacunarity were higher (P = .0005) and lower (P = .0014) in males, respectively.

CONCLUSION

WR.r was effective in revealing skeletonized bone trabeculae by removing non-trabecular noise. FD.r and FD.f revealed good agreement. FD.f, histogram parameters, and lacunarity differed based on dentate status, jaw location, and sex.