Investigation of dental cone-beam CT pixel data and a modified method for conversion to Hounsfield unit (HU).

OBJECTIVES

To investigate the relationship in dental cone-beam CT (CBCT) between the manufacturer-reported image pixel data and a modified conversion to CT number densities in Hounsfield unit (HU).

METHODS

A standardized CT phantom was imaged using typical clinical parameters on CBCT from three manufacturers (Carestream 9300, Carestream Health, Rochester, NY; J Morita 3D Accutomo, J. Morita Mfg. Corp., Kyoto, Japan; and Planmeca Promax 3D, Planmeca Helsinki, OY, Finland). Reconstructed axial slices were evaluated using regions of interest to ascertain the mean pixel value in five materials in the phantom. The Digital Imaging and Communications in Medicine data were also evaluated to determine if raw pixel data had been adjusted during the image reconstruction. A modified version of the existing manual HU conversion technique was applied, and the resultant slope and y-intercept were used to scale the pixel values ultimately to HU for all images.

RESULTS

The DICOM header data show that a modified rescale y-intercept was applied to both the Carestream and Planmeca image data yielding manufacturer-produced results in HU. The Morita pixel data were unmodified and report in shades of grey or grey values (GV). The Carestream manufacturer-derived HU measurements showed good correlation in air (-1000 HU), but all other materials ranged from 2.6 to 13.5 σ from the specified phantom value. Results in the modified conversion technique images fell within 1.0-2.4 σ from the specified phantom values.

CONCLUSIONS

While more studies are needed to test for regularity, this study suggests that our modified technique could be a means of getting more accurate quantitative data from dental CBCTs.