| Popis: |
The noise equivalent count (NEC) metric is designed to be proportional to the square of the average signal to noise in statistics-limited reconstructed PET images. It is equal to the trues count rate times the detective quantum efficiency averaged over the projection space region containing all lines-of-response traversing an object (typically a 20 cm diameter phantom). In list-mode PET, a number of attributes such as time and energy measurements are typically recorded in association with each independent event. Conventional PET signal processing, and the NEMA protocol for NEC measurement which corresponds to it, treats each independent event as contributing equally to NEC. Diversity combining signal processing provides an alternative, and in fact optimal, methodology which makes use of event attributes so as to maximize signal-to-noise. Conventional thresholding and windowing techniques may be replaced with adaptive, optimal weights, which may be derived from observed event and event ensemble characteristics. We discuss strategies for obtaining such scan-specific weights. The resulting signal-to-noise improvements can be used to shorten imaging times, decrease patient dose, or increase image quality. |
