Te. Southard et Ka. Southard, DETECTION OF SIMULATED OSTEOPOROSIS IN MAXILLAE USING RADIOGRAPHIC TEXTURE ANALYSIS, IEEE transactions on biomedical engineering, 43(2), 1996, pp. 123-132
An effective mass screening tool for detecting osteoporosis is current
ly lacking, Alveolar bone, routinely examined during periodic dental e
xaminations, may provide a window into the status of systemic bone den
sity, The primary objective of this investigation was to compare the p
erformance of various textural features, computed from dental radiogra
phs, in detecting early simulated osteoporosis of alveolar bone. Five
specimens of human maxillary alveolar bone were progressively decalcif
ied and the percentage calcium lost at each decalcification stage quan
tified, Two radiographs of each specimen, together with an aluminum st
epwedge, were exposed at 70 kVp at each stage, The test set of 140 rad
iographs was digitized, identical bony regions of interest selected fr
om the density-corrected images of each specimen, the regions digitall
y filtered to reduce film-grain noise, and textural features computed
on a line-to-line basis, Correlation analysis identified a set of feat
ures whose changes consistently exhibited a moderate-to-strong linear
association with bone mineral loss over a wide range of decalcificatio
n. Repeated measures analysis of variance was subsequently applied to
this set to measure the minimal decalcification that could be detected
by each feature under optimal conditions of x-ray beam angulation (0
degrees) and suboptimal conditions (+/-5 degrees). The best performing
features were mean intensity, gradient, Laws' texture energy measures
, and fractal dimension which detected 5.7 % bone decalcification at o
ptimal beam angulation and 9.4-12.6 % at suboptimal angulation.