Platform
Overview
The Hyperion PET detector platform provides a comprehensive set of hardware and software components to construct, scale, calibrate, and operate PET systems. While this makes the setup of a new PET scanner efficient and reliable, there is still the high flexibility that you need for research and for the adaptation to your specific layouts and applications.
Watch the video for a quick introduction to our PET detector platform!
Our Building Blocks
The MRI-compatible electronics platform consists of building blocks that can be combined for the construction of specific imaging devices. We can combine and adapt these building blocks for a multitude of PET scanner layouts, environments and required PET performance. Flexibility is the key to allow adaptions of the platform towards a multitude of imaging modalities, sensor readout or actor applications. From SPECT or gamma cameras to endless possibilities.
Detector Stack
The PET detectors are a stack of multiple layers, from scintillation crystals attached to a photo sensor array to the mechanical interface and the cable to the Singles Processing Unit.
Scintillation Crystals
Different scintillator topologies can be tailored to the applications. Whereas larger whole-body scanners may employ arrays made from 4mm x 4 mm x 19 mm LYSO crystals for good timing performance, an organ-specific PET scanner often focuses on a high spatial resolution. The Hypmed breast insert, for example, employs a three-layered DOI-capable scintillator made from 3425 crystals with a pitch of 1.33 mm.
Sensor Tile
The sensor tile is the electronic component of the detector stack containing the optical sensors. Currently, sensor tiles using analog SiPMs and the TOFPET2 ASIC from PETsys to digitize their signals are under development and will soon be available and used for the future PET and PET/MRI projects.
A previous implementation contains 6×6 DPC-3200 sensors from Philips Digital Photon Counting (PDPC), offering a total sensitive area of approx. 48×48 mm² with 144 channels and 36 time-to-digital-converters (TDCs). The readout electronics for the sensors are directly mounted on the rear side of the printed circuit board (PCB) to archive a relatively thin detector. An aluminum plate functions as mechanical and thermal interface to a combined mounting and cooling structure that allows precise mechanical positioning of the detector in a system geometry. The variety of sensor technology available today opens up the freedom to tailor the size and performance of the sensor tile to your application needs.
Tile Cable
The tile cable connects a detector stack to the main board of the electronics, the singles processing unit (SPU). The cables are custom-made and are available in two versions, allowing the cables to be attached from two different, opposite directions. The cable itself has a width of 6 mm and a height of 3 mm (or 12 mm x 1.5 mm when laid-oud flat). The minimum bending radius is 3.2 mm – yet the cable is ruggedized and up to 7000 bending cycles are allowed. Cable lengths of up to 5 m were tested successfully. A dual link allows a data rate of up to 1.8 GBit/s per detector stack.
Singles Processing Unit (SPU)
The current implementation of the SPU controls up to 15 detector tiles, collects and (pre-)processes the data, and sends it via optical 10-Gigabit Ethernet to the data acquisition system. The supply voltages for each tile can be controlled individually. As such, different detector technologies are supported. Multiple SPUs can be synchronized either galvanically or optically to ensure MRI compatibility.
Communication and processing on the SPU is currently implemented either on a cost-effective XC7K160T Kintex-7 FPGA (Xilinx) or on a XC7K410T offering more processing resources. The SPU transmits the data via a 10 GBit optical Ethernet link to the data acquisition system. A second optical transceiver can be used in parallel to double the data rate, or can be used to synchronize the SPUs of a PET system without a galvanic connection for highest MRI compatibility.
Data Acquisition and Processing Server (DAPS)
Sensor data and coincidence processing is realized in software which gives great flexibility to optimize for a specific application. The software runs on the Data Acquisition and Processing Server (DAPS) which collects the sensor data from all SPUs. The DAPS is optimized for performance: It can save the detector raw data for later processing and analysis, or directly process singles and coincidences on-line. The hardware can be tailored to the specific application needs and can range from a small desktop-like computer up to a powerful server.
Control Software
The PET system is controlled with a software offering an intuitive graphical user interface that communicates with the detector hardware, either directly with an Ethernet connection, or through the DAPS functioning as a router for control- and status information. The software reads the different PET scanner configurations from XML-files, configures the hardware, controls it during the scans, and displays status information graphically. Whereas regular users can operate the system with a few clicks, the expert user can change every setting of the scanner and even program scripts, e.g. to automate the acquisition of a series of scans. All data is saved with additional information and log files in a database to keep all scan information together.
Data Analysis Software
This software package is used for in-depth analysis of the raw detector data and allows to do the full processing from raw sensor data to singles and coincidences. At each processing step, data can be visualized in a variety of ways. Calibration values needed to correct raw sensor data, detectors and the full PET system can be generated and applied. Data integrity of the PET system can be checked and investigated for influences of a simultaneously operated MRI scanner. The latter can be done with a high temporal resolution that allows to investigate the PET data during the various stages of an MRI sequence. For example, the exact moment of when the gradient or RF system are switching their fields. The software is aimed at flexibility to allow for easy implementation of new processing algorithms. The computed calibration values are exported to the data acquisition and processing (DAPS) software with which the efficient on-line processing is performed.