Evaluation of optical fiber bundles for coupling a small LSO crystal array to a multi-channel PMT

One of the limiting factors in high resolution PET detectors is the readout of scintillation light from arrays of small crystals and the subsequent accurate identification of each array element. Furthermore, if position-sensitive or multichannel photomultiplier tubes (PMTs) are employed for readout, some form of light guide between the scintillator and PMT is required to maintain high packing fraction. The solution to this problem in microPET, a small animal PET scanner, was 1 to 1 coupling of the scintillation light from the crystal array to each channel of a multi-channel PMT using individual optical fibers. In developing a high resolution detector module for the next generation of microPET tomographs, the authors evaluated 8 types of optical fiber bundles with a 12×12 array of 1×1×10 mm LSO crystals. The authors examined the identification of array elements, energy resolution and light transmission. They tested 4 plastic and 4 glass fiber bundles, with various lengths and magnification factors, coupled to a Hamamatsu R5900-M64 multi-channel PMT. Although each fiber bundle resolved all the elements of the array, the best performance in terms of crystal separation (peak-to-valley ratio of 4:1), light transmission (85%) and energy resolution (~20%) was not achieved by any single sample. This is due to the trade-off between cross-talk and light transmission in the fiber bundle. The ultimate choice of fiber bundle therefore depends on the relative importance of light transmission (energy and timing resolution) and optical cross-talk (spatial resolution)