In radiotherapy with photon beams, the use of dynamic wedges, which ar
e obtained by the movement of one of the jaws, offers an increasing fl
exibility relative to the traditional use of metal wedges. But it is a
disadvantage for the measurement of absorbed dose distributions, beca
use the absorbed dose at each measurement point can only be obtained a
fter a complete movement of the jaw. Consequently, for radiotherapy pl
anning, an algorithm should be available that does not require measure
ments for any specific dynamically wedged beam, but is based on only a
modest number of measurements. In this paper, an algorithm for the ca
lculation of the dose distribution from dynamic wedges is described. T
his algorithm uses the convolution of pencil beam kernels with a non-u
niform field function. These pencil beam kernels are derived from empi
rical data resulting from measurements of the open beam only.