Evaluation of endodontic files in digital radiographs before and after employing three image processing algorithms

• Published in: Dento-maxillo-facial radiology 2004-01, Vol.33 (1), p.6-11
• Main Authors: Li, G, Sanderink, G C H, Welander, U, McDavid, W D, Näsström, K
• Format: Article
• Language: English
• Subjects: Algorithms; Area Under Curve; Dental Pulp Cavity - diagnostic imaging; Dentin - diagnostic imaging; Dentistry; Equipment Design; Humans; Image Processing, Computer-Assisted - methods; Image Processing, Computer-Assisted - statistics & numerical data; Medicin och hälsovetenskap; Phantoms, Imaging; Radiographic Image Enhancement - methods; Radiography, Dental, Digital - statistics & numerical data; ROC Curve; Root Canal Preparation - instrumentation; Tooth Apex - diagnostic imaging; Tooth Root - diagnostic imaging
• ISSN: 0250-832X; 1476-542X
• DOI: 10.1259/dmfr/26782200

Two digital image processing algorithms, one aimed at correction for exponential attenuation and one at correction for visual response, have been developed. The aims of the present study were to test whether digital radiographs processed with these algorithms improve determination of the length of endodontic files and whether such processed radiographs are comparable with the radiographs processed with a default image processing method employed by one commercially available digital intraoral system. A dried human skull embedded in an acrylic compound was used for exposing radiographs of the upper and lower premolars and molars with endodontic files (Kerr files size 10 and size 15) positioned to the full length of the roots or 1.5 mm short of it. Radiographs were then processed in three sets. In one set, the radiographs were processed to compensate for exponential attenuation and the response of the human visual system. In the second, the radiographs were processed with the same compensation but with an additional shift in grey levels so that the output luminance in dentin at root tips corresponds to the mean of the luminance range of a computer monitor. In the third, the radiographs were processed with the default processing method in the Sidexis program. Ten viewers evaluated all radiographs. Receiver operating characteristic (ROC) curves were obtained and areas under the curves were calculated. For file size 10, ROC curves for processed radiographs were higher than that for originals, while for file size 15, ROC curves for processed and original radiographs were close to each other. Significant differences were found between processed and original radiographs regarding areas under ROC curves for file size 10 but not between the differently processed radiographs. For file size 15, no significant differences were found. Radiographs processed to correct for attenuation and visual response may improve determination of the length of thin endodontic files. Such processed radiographs are comparable with the radiographs processed with the default processing method in the Sidexis program.