Untitled Document

(*) Summary.

Digital panoramic radiography is a diagnostic image technique which is increasing used today all over the world. Nevertheless, there is a relative lack of knowledge about the best compromise relationship between image quality and dose to the patient for these studies.

In the first project that we present twenty one panoramic images of an anthropomorphic phantom and 205 from patients were collected using a Kodak digital equipment. Tube current, beam energy and acquisition time were changed among studies to look for the best acquisition conditions which permit good image quality al low doses for patients. Kerma in air per area was measured as dose index. Image quality was graded using objective metrics as signal to noise ratio (SNR [dB]) and contrast to noise ratio (CNR), as well as visual evaluation with two expert observers. Reduction in dose was possible, reducing mAs mainly, without affecting image quality in a sensitive way. An optimized protocol for this equipment was also obtained for standard height and weight patients.

In the second project we present the evaluation of a new objective image quality metric, applied to dental images: HDR-VDP that was created to overcome poor correlation between objective and subjective digital image quality criteria. This metric has been tested successfully to analyze the effect of medical image compression over image quality. A study using 13 dental panoramic images of an anthropomorphic phantom was carried out. Images were acquired with digital Kodak equipment. Tube current and beam energy were changed among studies to analyze how image quality was affected with the acquisition parameter variation. The criterion of image quality of two expert observers was taken into account. Furthermore, HDR-VDP was fitted to the problem under analysis. Correlation between the metric and the subjective criterion was also analyzed. HDR-VDP had a correlation of r = 0.683 (p = 0. 013) with respect to the subjective criterion. With the metric calculated over 9 x 9 pixel windows in regions of interest we were able to detect differences in image quality for a typical range of kVp and mA in this type of study.

(**) Biodata.