Corrections for the effects of accidental coincidences, Compton scatter, and object size in positron emission mammography (PEM) imaging

Abstract

Positron emission mammography (PEM) has begun to show promise as an effective method for the detection of breast lesions. Due to its utilization of tumor-avid radiopharmaceuticals labeled with positron-emitting radionuclides, this technique may be especially useful in imaging of women with radiodense or fibrocystic breasts. While the use of these radiotracers affords PEM unique capabilities, it also introduces some limitations. Specifically, acceptance of accidental and Compton-scattered coincidence events can decrease lesion detectability. The authors studied the effect of accidental coincidence events on PEM images produced by the presence of /sup 18/F-Fluorodeoxyglucose in the organs of a subject using an anthropomorphic phantom. A delayed-coincidence technique was tested as a method for correcting PEM images for the occurrence of accidental events. Also, a Compton scatter correction algorithm designed specifically for PEM was developed and tested using a compressed breast phantom. Finally, the effect of object size on image counts and a correction for this effect were explored. The imager used in this study consisted of two PEM detector heads mounted 20 cm apart on a Lorad biopsy apparatus. The results demonstrated that a majority of the accidental coincidence events (/spl sim/80%) detected by this system were produced by radiotracer uptake in the adipose and muscle tissue of the torso. The presence of accidental coincidence events was shown to reduce lesion detectability. Much of this effect was eliminated by correction of the images utilizing estimates of accidental-coincidence contamination acquired with delayed coincidence circuitry built into the PEM system. The Compton scatter fraction for this system was /spl sim/14%. Utilization of a new scatter correction algorithm reduced the scatter fraction to /spl sim/1.5%. Finally, reduction of count recovery due to object size was measured and a correction to the data applied. Application of correction techniques for accidental coincidences, Compton scatter, and count loss due to image size increased target-to-background contrast ratios to approximately the maximum level theoretically achievable with this PEM system.