Positron Emission Tomography of bromine-76 and yttrium-86 results in the detection of coincident events that are not strictly associated with annihilation photon pairs. Instead, these coincidences occur because prompt gamma rays emitted by these nuclides result in cascades of photons that are emitted within the timing window of the PET scanner. Pairs of detected photons from these cascades are not angularly correlated and therefore contain little information regarding the location of their source. Furthermore, these coincidences are not removed by correction procedures (e.g., randoms, scatter) routinely applied to PET data. If left uncorrected, the cascade coincidences will result in spurious apparent activity within the PET images. A correction, applied within projection space, that removes the cascade coincidence signal from septa-in (i.e., two-dimensional) datasets is proposed and tested on phantom data.