首页> 外文会议>IEEE Nuclear Science Symposium Conference >Advanced Hardware Architecture for on-Line Data Acquisition in Clinical 3-D PET: Smart DRAM PCI Card for 14 M Event/sec Histogramming Across Terabytes of DRAM
【24h】

Advanced Hardware Architecture for on-Line Data Acquisition in Clinical 3-D PET: Smart DRAM PCI Card for 14 M Event/sec Histogramming Across Terabytes of DRAM

机译:在临床3-D PET中的在线数据采集的先进硬件架构:智能DRAM PCI卡14 M事件/秒直方图横跨DRAM TERABYTES

获取原文

摘要

Clinical PET places a demanding burden on the data acquisition architecture - especially when using larger, higher sensitivity detector arrays coupled with demands for high patient throughput. Pure list-mode data acquisition can have limited acceptance due to either too-slow event-packet collection speeds and/or too-slow post-acquisition processing. The typical resolution for this problem has been to perform both LOR-to-bin mapping (rebinning) and histogramming (+/-unity to memory) in an on-line manner - i.e. projection space "smograms" are generated during the acquisition while the patient is still in the tomograph. This article describes a new design for very-high-speed online histogramming hardware - the Smart DRAM PCI card. This hardware is significant in the quest to address the needs of fast on-line histogramming into large DRAM arrays. This card has an integral dual-Fibre-Channel interface for receiving (and retransmitting) PET bin-address packets. Also, the card has on-board quad-interleaved ECC DDR SDRAM (4 DIMMs & up to 4 G bytes total per card) with a demonstrated capacity for sustained histogramming rates in excess of 14 M events/sec. The retransmit-packets-on-Fibre-Channel feature supports a daisy chain of multiple such Smart DRAM cards. This means the architecture can easily make terabytes of memory available for on-line generation of projection data. Support is planned for time-of-flight and weighted histogramming as well as online gating.
机译:临床宠物对数据采集架构的要求苛刻 - 特别是在使用更大的更高的敏感性探测器阵列时加上高病毒吞吐量的需求。纯粹的列表模式数据采集可能由于太慢的事件 - 包收集速度和/或过于缓慢的采集后处理而具有有限的接受。以在线方式执行该问题的典型分辨率,以便在线方式执行LOR-to-bin映射(repinning)和直方图(+/- unity) - 即在获取期间生成投影空间“SMACHS”病人仍然在Tomprox中。本文介绍了一款非常高速在线直方图硬件的新设计 - 智能DRAM PCI卡。此硬件在寻求将快速在线直方图的需求中解决大型DRAM阵列的需要很重要。该卡具有一体的双光纤通道接口,用于接收(和重传)PET宾寄存器数据包。此外,该卡具有车载Quad交错的ECC DDR SDRAM(每张卡片4个DIMM,最高可达4 G字节),持续的直方图率超过14米的持续直方图/秒。 Retransmit-Packets-on-FibreS通道功能支持多个此类智能DRAM卡的菊花链。这意味着架构可以很容易地制作可用于投影数据的在线生成的Tberabyte。计划用于飞行时间和加权直方图以及在线门控。

著录项

相似文献

  • 外文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号