Multiplexing strategies for cMiCE PET detectors

Continuous miniature crystal element (cMiCE) PET detectors use monolithic scintillators coupled to arrays of photosensor elements and statistics based methods for positioning of detected events. Current implementations acquire and utilize all photosensor array channels for event positioning (e.g., 64 channels for an 8×8 PMT or SiPM array). We investigate different multiplexing strategies to reduce the number of acquired signal channels and their impact on positioning performance. This study was conducted using data collected from a cMiCE PET detector. Sixty-four signals were collected per event and data were binned into four depth of interaction regions. The multiplexing strategies were implemented in software. Multiplexing strategies investigated included rowcolumn (RC) summing of signals (64 channel -> 16 channel); sampling based upon a modulus 3 and modulus 5 patterns of detector channels (64 -> 16); variants of RC summing (e.g., 64 -> 19 or 64 -> 8); and multiplexing based upon principal component analysis. The average X,Y intrinsic spatial resolution for the cMiCE detector using all 64 channels for positioning was 1.26 mm FWHM in X and Y. For standard RC summing of signals the average intrinsic X,Y spatial resolution was 1.32 mm. The intrinsic spatial resolution for the modulus 3 and 5 multiplexing was significantly worse at 1.43 mm FWHM. A RC summing method that used three additional multiplexed channels along the edges of the crystal provided similar decoding performance as standard RC summing (i.e., 1.31 mm) but better visual spatial positioning in the corners and edges of the detector. However, the most encouraging results were using multiplexing methods based upon the principal components of the detector signals; the intrinsic spatial resolution for this method was best of all the multiplexing methods (i.e., 1.30 mm FWHM) and it proved to be fairly robust to slight changes in the weighting factors. In conclusion, signal multiplexing techniques can be applied to monolithic crystal PET detectors that utilize statistics-based positioning methods. Reductions in the number of acquisition signal channels of a factor of 3-5 resulted in only 4-10% degradation in spatial resolution performance.