A prototype PET scanner with DOI-encoding detectors

Abstract

Detectors with depth-encoding allow a PET scanner to simultaneously achieve high sensitivity and high spatial resolution.

Methods: A prototype PET scanner, consisting of depth-encoding detectors constructed by dual-ended readout of lutetium oxyorthosilicate (LSO) arrays with 2 position-sensitive avalanche photodiodes (PSAPDs), was developed. The scanner comprised 2 detector plates, each with 4 detector modules, and the LSO arrays consisted of 7 x 7 elements, with a crystal size of 0.9225 x 0.9225 x 20 mm and a pitch of 1.0 mm. The active area of the PSAPDs was 8 x 8 mm. The performance of individual detector modules was characterized. A line-source phantom and a hot-rod phantom were imaged on the prototype scanner in 2 different scanner configurations. The images were reconstructed using 20, 10, 5, 2, and 1 depth-of-interaction (DOI) bins to demonstrate the effects of DOI resolution on reconstructed image resolution and visual image quality.

Results: The flood histograms measured from the sum of both PSAPD signals were only weakly depth-dependent, and excellent crystal identification was obtained at all depths. The flood histograms improved as the detector temperature decreased. DOI resolution and energy resolution improved significantly as the temperature decreased from 20 degrees C to 10 degrees C but improved only slightly with a subsequent temperature decrease to 0 degrees C. A full width at half maximum (FWHM) DOI resolution of 2 mm and an FWHM energy resolution of 15% were obtained at a temperature of 10 degrees C. Phantom studies showed that DOI measurements significantly improved the reconstructed image resolution. In the first scanner configuration (parallel detector planes), the image resolution at the center of the field of view was 0.9-mm FWHM with 20 DOI bins and 1.6-mm FWHM with 1 DOI bin. In the second scanner configuration (detector planes at a 40 degrees angle), the image resolution at the center of the field of view was 1.0-mm FWHM with 20 DOI bins and was not measurable when using only 1 bin.

Conclusion: PET scanners based on this detector design offer the prospect of high and uniform spatial resolution (crystal size, approximately 1 mm; DOI resolution, approximately 2 mm), high sensitivity (20-mm-thick detectors), and compact size (DOI encoding permits detectors to be tightly packed around the subject and minimizes number of detectors needed).

FIGURE 1

A schematic view of the prototype PET scanner. Top: configuration 1, bottom: configuration 2.

FIGURE 2

Flood histograms as a function of irradiation depth calculated by using position-encoding energy signals from PSAPD 1, PSAPD 2 and both PSAPDs.

FIGURE 3

Flood histograms of the whole array measured at three temperatures.

FIGURE 4

Dependence of DOI resolution (FWHM) on depth. The results are the average of all crystals and obtained from the DOI profiles of individual crystals.

FIGURE 5

Timing spectra of three different crystals and all 25 crystals in a 5×5 LSO array. Note that there is a time shift in the spectra with the largest difference being between a central crystal and a corner crystal.

FIGURE 6

Flood histograms and DOI responses of all eight detectors obtained from a phantom scan.

FIGURE 7

Images of the line source phantom and the hot rod phantom reconstructed with 20, 10, 5, 2 and 1 DOI bins respectively. The images were acquired with two different scanner configurations as shown in Figure 1. The images were reconstructed by filtered backprojection with Shepp-Logan filter cut off at the Nyquist frequency for the line source phantom and 0.5 of the Nyquist frequency for the hot rod phantom. The rod diameters in the hot rod phantom are 0.75, 1.0, 1.35, 1.70, 2.0 and 2.4 mm with a rod-to-rod distance of twice the rod diameter. Note the improvement in the sharpness of the reconstructed images with increasing number of DOI bins.