Preliminary results have been obtained with a non-linear filter system for projection data in computed tomography imaging. The non-linear filter system consists of a sequence of three filters: a non-linear median smoother, a linear smoother and a non-linear correction filter. The non-linear correction filter is a modified Kalman filter. Both patient data and simulated phantom data have been used in evaluating and designing the non-linear filter system. There are six parameters to be adjusted to the type of projection data considered. Additional work is required to optimize the filter system. Our preliminary results show a significant improvement in image resolution for a noise reduction comparable to what is obtained with conventional linear filtering techniques. However, this is at the cost of enhancing the streak artifacts.

Date Published: 15 November 1978
PDF: 7 pages
Proc. SPIE 0152, Recent and Future Developments in Medical Imaging I, (15 November 1978); doi: 10.1117/12.938189

Published in SPIE Proceedings Vol. 0152:
Recent and Future Developments in Medical Imaging I
Norman A. Baily, Editor(s)