Estimating random coincidences from singles count rates for improved RatCAP image accuracy

The delayed coincidence window technique has been established as the standard method to estimate random coincidences in PET. It has successfully been incorporated into the RatCAP’s singles list mode data processing scheme, and has been demonstrated to be accurate to ≫99%. However, similar to prompt coincidences, the randoms estimate obtained using this technique is subject to Poisson noise which propagates into the reconstructed image. This limits administered activities at high count rates since the randoms rate is proportional to the square of the count rate. The RatCAP, a 3D small animal tomograph is attached to the head of the rat, making it susceptible to random coincidences both from within and outside the field of view. The randoms fraction in recent 11C-raclopride studies with the current timing window have been observed to be up to 50%, making randoms correction a vital step prior to reconstruction. In this paper, we investigate a method that generates randoms estimates with higher statistical quality using singles count rates. Details of implementation will be discussed, and the improvement in the noise equivalent count rate will be validated.