Evaluation of geometrical arrangements of high resolution sensors in PET probe configuration

The probe arrangement in positron emission tomography (PET) is a combination of a high resolution detector (probe) placed within a whole body PET scanner. An additional subset of high-resolution data is recorded that contains events where one of the annihilation photon pair interacted in the high resolution detector. A study was designed to evaluate impact of probe geometry and efficiency on the quality of obtained image. The performance was predicted by estimating mean and variance of efficient estimators such as MLEM based on imaging matrix evaluated at particular probe geometry. Results were compared to data measured at a high-resolution PET probe prototype constructed at University of Michigan. The prototype consists of two arcs of standard PET detector elements at 50 cm radius combined with silicon sensors segmented into 1 mm
³
at 5.7 cm radius. A high-resolution Derenzo phantom filled with
¹⁸
F-flourodeoxyglucose was rotated during a scan. Limited angle probe geometry was simulated by selection of a subset of phantom angular views. Data reconstruction was matched to algorithms evaluated in theoretical predictions. Two parameters, contrast recovery coefficient (CRC) and signal to noise ratio (SNR) were compared to predicted values. The data shows a 50 % increase in CRC and SNR at an event ratio of 10 % where improvement is maximized for a subset of angular positions. Splitting high-resolution data to two perpendicular angular positions improves CRC and SNR.