ABSTRACT
In conventional 3D-CT image processing, the images are influenced by subjective threshold settings. The purpose of this study was to evaluate an objective threshold setting technique based on the discriminant analysis method. The concentration of contrast medium in a joint of a simulated upper cervical spine phantom was changed, and its threshold was measured from scanned data by using the discriminant analysis method, and mean CT attenuation was measured. On the other hand, an accurate image of the corresponding joint in the phantom was made, and its minimum threshold was measured. Regression analysis between the adjusted minimum threshold and mean CT attenuation of the region of contrast medium was performed. The obtained linear regression formula was applied to the threshold settings in five cases for atlanto-axial 3D-CT facet arthrogram(3D-CTF), and the accuracy of the images was examined. There was a strong correlation between the adjusted threshold and mean CT attenuation, and the obtained linear regression formula was y=0.625x-141(r2=0.991, p<0.01). This equation could be used clinically for correction of the threshold settings. We propose the following method for threshold setting of 3D-CTF: the threshold of the region of contrast medium is measured using the discriminant analysis method, then the adjusted minimum threshold for the threshold settings of 3D-CTF is calculated from mean CT attenuation. The method described herein is an objective, general-purpose methodology that is applicable to various types of 3D-CT.
