Digital radiography of interproximal bone loss; validity of different filters

The aim of the present study was to compare linear measurements of interpro ximal bone loss on digitized radiographic images after application of diffe rent filters to the gold standard of intrasurgical measurements. Immediatel y before surgery 50 radiographs of 50 periodontally-diseased teeth exhibiti ng interproximal bone loss, were obtained by a standardized technique in 33 patients. Intrasurgically, the distances from the cementoenamel junction ( CEJ) to the alveolar crest (AC) and from the CEJ to the deepest extension o f the bony defect (BD) were assessed. All radiographs were digitized with a flatbed scanner (resolution: 600x1200 dpi). Using the FRIACOM-soft-ware, t he linear distances CEJ to AC and CEJ to ED were measured at 50 interproxim al lesions from the digitized but unchanged radiographic images and also af ter use of 7 different basic image-processing modes (enhancement of contras t [dynamics], inversion, high-pass, enhancement of gray-level differences, mean value, histogram correction, spreading of grey values) with 11-fold ma gnification. Neither the measurement of the distance CEJ to AC on the uncha nged images nor assessments with any of the filters revealed significant di fferences from the gold standard. Multivariate analysis of variance showed the respective filter (p=0.009), intrasurgical and radiographic assessments (p<0.0001), to statistically significantly, influence the difference betwe en intrasurgical and radiographic measurements of the distance CEJ to ED. T he underestimation of the intra-surgically assessed distance CEJ to ED by r adiographic measurements ranged from 0.3+/-2.0 to 0.8+/-1.9 mm. The filter "mean value" underestimated interproximal bone loss statistically significa ntly more than the digitized but unchanged radiograph (p=0.05). In this stu dy, basic digital manipulations (filters) of radiographic images failed to result in statistically significantly more valid measurements of interproxi mal bone loss when compared to the unchanged but digitized images. All radi ographic assessments on the digitized images except for use of one filter ( mean value) came close to the intrasurgical gold standard.