Abstract: An apparatus for reducing image artifacts caused by over-ranging or clipping of the data collected in a tomographic scan fits a geometric model to the in-range data of each projection. The geometric model is sized to the in-range data by summing all of the data of the projection to obtain a value of the total slice volume. This geometric model is used to compute the slope of the over-range data from the last point of in-range data and this extrapolated data is substituted for the over-range data. The correction process is implemented in pipeline form by convolving an over-range mask indicating which data is in-range and which data is over-range with a box car convolution kernel which produces a trapezoidal correction mask. This correction mask, multiplied by the uncorrected projection data provides the appropriate slope to its clipped portions.