| 摘要 |
An algorithmic method is used for suppressing artifacts in computed tomography raw data, on the basis of the determination and subsequent subtraction of a correction sinogram from a measured starting sinogram. The method includes high-pass filtering of a starting sinogram in the channel direction, and low-pass filtering in the projection direction in order to improve the signal-to-noise ratio. Thereafter, the magnitude of a weighted gradient of each data point in the low-pass-filtered sinogram is formed, both in the projection direction and symmetrically about the corresponding channel axis. The data point is eliminated if the change amplitude thereof exceeds a first defined threshold value. Residual data points are removed in the low-pass-filtered sinogram if their amplitude exceeds a second defined threshold value, and low-pass filtering of the resulting sinogram occurs in the form of averaging in the projection direction. The correction sinogram thus obtained is finally subtracted from the starting sinogram.
|
