Professional Documents
Culture Documents
Linearity Energy and Uniformity Calibration
Linearity Energy and Uniformity Calibration
Linearity calibration
Purpose: The Anger calculation gives strongly nonlinear image. The nonlinearity can be
corrected by generating a vector field and so correct the calculated position.
Procedure: With Tc-99m applying the linearity lead masks (parallel 9mm equidistance
1mm slits in X or Y direction) linearity images have to be taken. Then using the
LinCALC program (written by Sándor Török) one can calculate linearity correction
tables (containing the correction vector field components) in an iterative way. The
method uses the X and Y images. Knowing the real positions of the lines one can
calculate the required shifts for the given direction – in case of X linearity image,
linearity can be calculated with a density of 16,384 pixels but only at 9mm steps for Y
direction. Since a more dense grid is required for correction (65,536*65,536), internal,
not measured grid points are interpolated linearly. After correction is calculated, linearity
coefficients (see NEMA quantities) is calculated on the acquired and corrected images. If
it is not satisfactory, these images can be saved and the linearity correction table can be
made more precise using the saved results for recalculation the tables. This iteration has
to be performed until acceptable linearity coefficients are achieved.
When linearity correction is turned on, the position calculation is
16),(__PosYPosXPosXmmppixmmPosXLxΛ+⋅=
and similarly for Y direction and Λ represents the x component of the linearity
x
correction vector field.
Reference: not available. All information comes from discussing with Sándor Török.
Note: Linearity correction intrinsically includes a new XY factor calibration, since the
imaged lines are adjusted not to be straight but being equidistance according to the lead
mask. Therefore X_factor and Y_factor should be determined from the fact that PosX/16
needs to be within the range [0, 1024]. The precise scaling will be performed by means of
linearity calibration anyway.