Measurement of CT Resolution in Three Dimensions

David O. Hearshen, David A. Reimann, and Michael J. Flynn

Radiological Society of North America, 81st Scientific Assembly and Annual Meeting of the Radiologic Society of North America, Chicago, Illinois, 26 November-1 December. 1995. Radiology, 197(P Supplement):291, November 1995.

PURPOSE: Quantitative measurement of CT resolution in three dimensions using a simple phantom and one acquisition.

MATERIALS AND METHODS: A 35 mm plastic sphere is imaged with sequential slices. The precise center of the sphere is iteratively determined by minimizing directional surface response differences. The distance from the sphere's center to every voxel is computed. Gray level as a function of distance from the sphere's center is binned at 0.1 pixel spacings. The resulting spread function is smoothed and differentiated to yield a resolution spread function. Because many phases of the sphere surface and the voxel structure are present, the resulting measure does not contain the effects of the final image digitization process. Directional as well as average 3D resolution can be measured.

RESULTS: Changes in imaging parameters such as focal spot size, convolution filter, and slice thickness yielded expected variations in resolution. Preliminary results from separate data obtained with 2 focal spot sizes on a GE HiSpeed scanner using a thin slice protocol indicated the method can distinguish between large/small focal spot sizes which produce a 4 percent difference in transaxial resolution (1.31,1.37 mm FWHM) and a 25 percent difference in axial resolution (1.17,1.46 mm FWHM).

CONCLUSION: A fast quantitative measure of the presampled axial and transaxial resolution spread function resolution has been developed. The large number of points used over regions of the sphere provides excellent statistical precision.

Copyright © 1995, David A. Reimann. All rights reserved.