The β-CUBE, a high-end compact preclinical benchtop PET for total body imaging (SNMMI, 2017)

Authors: Mollet Pieter 1, Deprez Karel 1, Vandeghinste Bert 1, Neyt Sara1, Marcinkowski Radoslaw2, Vandenberghe Stefaan2, Van Holen Roel2 Affiliations


  1. MOLECUBES NV, Ghent Belgium.
  2. IBiTech – MEDISIP – INFINITY, Ghent University, Belgium.


To further improve preclinical PET, new systems should improve spatial resolution and sensitivity. Ideally this is obtained without increasing the total cost of the system. Based on monolithic scintillators with highly sampled readout and Depth-Of-Interaction capability, MOLECUBES developed a compact preclinical PET system, called the β-CUBE. Here, we report on the design, the specifications, performance evaluation and in-vivo imaging results.


The system consists of a ring of 9 detector modules with an inner diameter of 76 mm and a long axial FOV of 13.3 cm, enabling total body mouse and rat imaging. Each module contains five 25.4 x 25.4 x 8 mm thick monolithic LYSO scintillators coupled to an analogue SiPM (Hamamatsu MPPC). Peltier coolers control the SiPM temperature at a stable 23 degree Celsius. Dedicated read-out electronics read the signal of each pixel of the MPPC. All detector units are connected to 5 coincidence boards. The acquisition and reconstruction servers are integrated inside the 54 cm x 54 cm x 54 cm system. The acquisition server performs on-the-fly GPU-based statistical event positioning of acquired pixel signals. The reconstruction server runs a GPU-based Tube of Response-based OSEM reconstruction and makes the final images available to the user in DICOM format. The attenuation map for the PET image reconstruction is obtained from the previously introduced X-cube, providing sub-minute microCT-imaging at low dose (less then 6 mGy for the fastest whole body mouse scan). Both systems are controlled wirelessly using a tablet or a computer with a single integrated software. Sequential PET/CT is acquired with a single animal bed shared by both systems while co-registration is performed automatically.

The resolution in image space is further improved by using continuous Depth Of Interaction functions. All reconstructions were done with 50 iterations of our OSEM reconstruction. The PET detector intrinsic resolution was measured with a 0.4 mm collimated beam source. Spatial resolution is evaluated using point sources and a hot-rod phantom and image uniformity and recovery with a NEMA image quality phantom.


The intrinsic detector resolution is 0.76 mm full width at half maximum (FWHM). The reconstructed tangential and axial image spatial resolution is 0.874 mm FWHM and 0.780 mm FWHM respectively at the center of the field of view (CFOV). Radial spatial resolution was nearly constant as a function of the radial offset, varying from 0.9 mm FWHM at a radial distance of 10 mm from the center to 1.15 mm FWHM at a distance of 25 mm. Using a hot-rod phantom, the 0.8 mm rods are clearly visible while some of the 0.7 mm rods can be discriminated. Uniformity was 16.6% and the hot rod recovery coefficients ranged from 0.28 to 0.96. The absolute peak sensitivity in the CFOV is 10.2 %. In-vivo scans of a mouse and a rat using both NaF-F18 and F18-FDG shows that the resolution and sensitivity of our system is exquisite over the entire FOV and has large potential for quantitative and dynamic imaging.


The design of the β-CUBE resulted in a truly high-end PET imaging system for rodents. The compact and portable system can be installed in any certified environment on a standard lab table and allows for stationary total body mouse imaging with sub-millimeter resolution in combination with high sensitivity.