A novel single-ended readout depth-of-interaction PET detector fabricated using sub-surface laser engraving

We propose a novel scintillation detector design for positron emission tomography (PET), which has depth of interaction (DOI) capability and uses a single-ended readout scheme. The DOI detector contains a pair of crystal bars segmented using sub-surface laser engraving (SSLE). The two crystal bars are optically coupled to each other at their top segments and are coupled to two photo-sensors at their bottom segments. Initially, we evaluated the performance of different designs of single crystal bars coupled to photomultiplier tubes at both ends. We found that segmentation by SSLE results in superior performance compared to the conventional method. As the next step, we constructed a crystal unit composed of a 3  ×  3  ×  20 mm³ crystal bar pair, with each bar containing four layers segmented using the SSLE. We measured the DOI performance by changing the optical conditions for the crystal unit. Based on the experimental results, we then assessed the detector performance in terms of the DOI capability by evaluating the position error, energy resolution, and light collection efficiency for various crystal unit designs with different bar sizes and a different number of layers (four to seven layers). DOI encoding with small position error was achieved for crystal units composed of a 3  ×  3  ×  20 mm³ LYSO bar pair having up to seven layers, and with those composed of a 2  ×  2  ×  20 mm³ LYSO bar pair having up to six layers. The energy resolution of the segment in the seven-layer 3  ×  3  ×  20 mm³ crystal bar pair was 9.3%-15.5% for 662 keV gamma-rays, where the segments closer to the photo-sensors provided better energy resolution. SSLE provides high geometrical accuracy at low production cost due to the simplicity of the crystal assembly. Therefore, the proposed DOI detector is expected to be an attractive choice for practical small-bore PET systems dedicated to imaging of the brain, breast, and small animals.