Correction algorithms for signal reduction in insensitive areas of a small gamma camera

Abstract

The radiation sensitive part of the wide-field, small gamma camera Sentinella 108 consists of a monolithic, 102 x 102 mm2 wide CsI(Na) crystal coupled to four position-sensitive photomultipliers. These are read out via a resistor mesh reducing the 256 PMT channels to four digitized values which allow for a simple image reconstruction, based on polynomials, with high spatial resolution. The sensor areas between neighboring PMTs suffer from reduced detection efficiency for scintillation light. The corresponding lack of signal amplitude translates to a poor uniformity and additional image compression in the border regions of the images after the first step of reconstruction. These effects cannot be completely eliminated by a global calibration procedure alone. We have developed and evaluated two classes of algorithms to compensate for this reduction of signal amplitude. Corrections are applied for each detected event separately, maintaining the real-time capability of the gamma camera. Algorithm 1 is based on a four-parameter, geometric calculation of the four output signals. This model aims to attack the underlying lack of detection efficiency without further assumptions on the true event distribution, preserving the global polynomial image decompression. Algorithm 2, to the contrary, starts from measured reference distributions of point sources which are used to model local deformations and energy loss without further information on the underlying detector geometry. Here too, a good image quality has been obtained after application of the deduced corrections. These algorithms are applied in the first step of event reconstruction. Within an energy interval of +-0.2 E_p around the photopeak position of 99m-Tc, E_p = 140.5 keV, and comparing several cameras, global uniformities around 5-8% have been obtained with both models. Measured energy resolutions are about 9-14%.