Skip over global navigation links

High Resolution Research Tomograph

Brief Description

The Motion-compensation OSEM* List-mode Algorithm for Resolution-recovery Reconstruction (MOLAR) system is the result of an ongoing collaboration between three organizations in the NIH Intramural Research Program as well as Yale University, SUNY Buffalo, and CPS Innovations, Knoxville, TN, USA. It is a complete system for managing and performing high-resolution, iterative reconstructions of positron emission tomography (PET) data. MOLAR has been designed for use with the ECAT HRRT (High Resolution Research Tomograph, CPS Innovations) operating in list-mode. Due to the pluggable component design of the software, however, the MOLAR reconstruction engine is readily adaptable to any PET scanner, including frame-mode scanners. Reconstructions are performed on a parallel cluster of commodity computers. One of the goals of the project is to provide complete reference implementations (i.e., with physical effects incorporated into the model) of common iterative reconstruction algorithms such as OSEM*. Another goal of the project is to provide to the PET research community a general software framework for performing list-mode or frame-mode reconstructions on the HRRT or any other PET scanner. The framework has been designed to allow collaborating groups or individuals the opportunity to contribute their own components.

* OSEM: Ordered Subsets Expectation Maximization reconstruction algorithm.

Front view of the ECAT HRRT scanner by CPS Innovations, Knoxville, TN. The topology of the HRRT is an octagon of dual-layer (LSO/LYSO) detector banks, each bank being an array of 9*13 blocks, and each block containing 8*8 crystals of size 2.1*2.1*10-mm each. With 119,808 crystals, the HRRT has 4.5 billion potential lines of response. The HRRT has a 35-cm patient port, making it suitable for human brain studies as well as large animal studies. The massive amount of data generated by the HRRT, coupled with interest in high-resolution, high-sensitivity reconstructions, motivates our present work.
Front view of the ECAT HRRT scanner by CPS Innovations, Knoxville, TN. The topology of the HRRT is an octagon of dual-layer (LSO/LYSO) detector banks, each bank being an array of 9*13 blocks, and each block containing 8*8 crystals of size 2.1*2.1*10-mm each. With 119,808 crystals, the HRRT has 4.5 billion potential lines of response. The HRRT has a 35-cm patient port, making it suitable for human brain studies as well as large animal studies. The massive amount of data generated by the HRRT, coupled with interest in high-resolution, high-sensitivity reconstructions, motivates our present work.

One slice of a sinogram, binned by MOLAR, of a uniformity cylinder phantom. The diamond patterns are due to the detector gaps.
One slice of a sinogram, binned by MOLAR, of a uniformity cylinder phantom. The diamond patterns are due to the detector gaps.

Central transverse slice of the global sensitivity image generated by MOLAR with 50-M randomized events, without attenuation or motion correction. The spoke-and-wheel pattern is a result of gap-effect cancellation along the lines of response that connect two gap regions.
Central transverse slice of the global sensitivity image generated by MOLAR with 50-M randomized events, without attenuation or motion correction. The spoke-and-wheel pattern is a result of gap-effect cancellation along the lines of response that connect two gap regions.
Central transverse slice of the global sensitivity image generated by MOLAR with 50-M randomized events, without attenuation or motion correction. The spoke-and-wheel pattern is a result of gap-effect cancellation along the lines of response that connect two gap regions.

Recent Accomplishments

Progress was made in improving the quality of the reconstruction software as well as in extending the reach of the HRRT scanner at NIH. The reconstruction engine has gone into production mode, reconstructing human brain images on a routine basis. As such, consideration in being given to splitting the computing cluster into a production cluster and a development cluster. Discussions are underway on how and where to host the production cluster. Significant progress was made on a number of the physical corrections that are modeled in the reconstruction software, most notably the randoms estimation, the scatter correction, and the correction for motion using the Polaris system.

In FY 2006, much was done to improve the quality of the reconstruction software. Extensive improvements were made in the logging of reconstruction jobs to provide useful monitoring and diagnostic information to users. One of the features of the reconstruction engine is the “swapping” of large event lists into and out of memory; the software that controls the swapping has been improved for efficiency as well as reliability. The software build process has been simplified and documentation has been provided. The validation of the parameter file has been completely re-engineered. A canonical set of tests have been established as well as procedures for testing new releases of the software. Error handling and reporting has been refactored throughout the code. Procedures have been established for performance testing and platform validation.

Animal studies were performed on a variety of novel tracers including 18F-FCWAY (serotonin 5-HT1A antagonist), 18F-FDOPA (dopamine synthesis), 11C-Raclopride (dopamine D2 receptor antagonist), 11C-Leucine (protein synthesis), 11C-DASB (serotonin transporter), 18F-mGluR5 (metabolic glutamate receptor), 11C-PBR (peripheral benzodiazepine receptors), 11C-Rolipram (phosphodiesterase 4 inhibitor), 11C-CB1 (canabinoid receptors), and 11C-RWAY (serotonin 5-HT1A antagonist). Human studies were performed on 11C-PBR, 11C-Ropipram, 11C-Leucine, and 18F-FDOPA by three investigators in NIMH and NINDS.

Current and Future Work

Now that a quality baseline for the software has been establish, progress on new features can be expedited. One of the important new features to be implemented is a “faster” reconstruction that does not include full resolution recovery. Significant testing is required in order to establish parameters for performing these faster reconstructions. A better understanding is needed into the types of artifacts that are generated by taking the “shortcuts” necessary to achieve the faster performance. Research into the potential for new reconstruction algorithms under certain situations (such as cold spot recovery) may also be attempted in FY 2007. There is currently considerable interest in estimation of kinetic parameters in 4-dimensional reconstructions. Time permitting, such investigation could be initiated here.

Over the past year, a few bugs were observed on an intermittent basis. These bugs include the appearance of NaNs (not a number) in the reconstruction logs due to domain errors. Another bug is the occasional appearance of a “photon torpedo”, a bright comet-like streak that seems to travel between slices in the reconstruction. In FY 2007 we hope to resolve these bugs.

Collaborators

  • Robert Innis, M.D., Ph.D., Chief, Laboratory of Molecular Imaging, NIMH
  • Jeih-San Liow, Ph.D., Laboratory of Molecular Imaging, NIMH
  • David Goldstein, M.D., Ph.D., Chief, Clinical Neurocardiology Section, NINDS
  • Peter Herscovitch, M.D., Director, Positron Emission Tomography Department, Clinical Center
  • Craig Barker, Ph.D., Positron Emission Tomography Department, Clinical Center
  • Shanthalaxmi Thada, Positron Emission Tomography Department, Clinical Center
  • Richard Carson, Ph.D., Yale University
  • Mario Rodriguez, Ph.D., Yale University
  • Rutao Yao, Ph.D., State University of New York, Buffalo

Publications

M. Rodriguez, J.-S. Liow, S. Thada, M. Sibomana, S. Chelikani, T. Mulnix, C.A. Johnson, C. Michel, W.C. Barker, and R.E. Carson, “Count-Rate Dependent Component-Based Normalization for the HRRT,” submitted to the IEEE Transaction on Nuclear Science.

Metrics

All metrics represent aggregates for 2006

  • Number of users in: 8
  • Number of successful reconstructions: 4714
  • Average number of processors per reconstruction: 14
  • Average wall-clock time per reconstruction: 4.04 hours
  • Cluster utilization 85%

Up to Top

This page last reviewed: November 09, 2011