C.A.S.I.S. Workshop 2003 Abstract Proceedings

Conference · Mon Nov 10 00:00:00 EST 2003

OSTI ID:15013615

Azevedo, S

Thirty five years ago, while in the neutron cross section group led by Robert Howerton at LLNL, the concept of reconstructing a three dimensional spatial distribution from its projections onto two dimensional planes was tackled by some of us using three now well known methods: simple back projection, Fourier projection theorem methods and iterative least squares algebraic reconstruction. The method of iterative least squares reconstruction was implemented on patient data in the early 1970s using photons from radionuclides detected by the Anger Camera. The method useful for computed tomography was modified to include the attenuation of the photons from an unknown source through an unknown attenuation distribution (a problem thought to be intractable until 1974). These methods along with a multitude of other methods developed by my small group of Ronald Huesman, Grant Gullberg, William Greenberg and Stephen Derenzo were prepared as a library with examples in FORTRAN, RECLBL. Those codes were found useful for computed tomography, geophysical problems and plasma confinement research topics in addition to their use in Nuclear Medicine. The codes were used even in the early days of magnetic resonance imaging when back projection of filtered projection data were used before the incorporation of phase encoding methods. In 1970s computed tomography of a single section of the brain required 4 minutes for single photon tomography or positron emission tomography about 30 minutes were required. Even proton and helium ion tomography were accomplished in the 1970s but with more that 2 hours of data acquisition. Thirty years later CT systems deliver 16 sections per second with 1 mm resolution and PET systems acquire 40 sections with about 4 mm resolution in 5 minutes. Computation times have reduced from 18 hours on the CDC 6600s and 7600s for gated, list mode data to less than 5 minutes in the last 30 years. Similar ratios of improvement have benefited ultrasound and magnetic resonance imaging. So what are the constant elements of imaging and what are the elements which change in the future? By the constant elements I refer to those principles which are time or history invariant (e.g., Fourier projection theorem, methods for detection of single or limited number of sources are not optimal for continuous distributions; limited angle or incomplete data are missing sources, which cannot be set to zero; back projecting data into unnatural pixels or voxels will result in erroneous reconstructions, etc.). The elements, which will change in the future include improvements in 5 attributes of an imaging modality: spatial resolution, time resolution, volume of the imaged object, signal to noise, contrast sensitivity. Signal to noise improvements include methods to reduce scatter in photon imaging as well as improvement in data transmission and processing before digitization. The one attribute of most importance in the selection of the imaging modality to solve a problem (e.g., electric source imaging vs electron density imaging) is contrast sensitivity because it is this property which distinguishes one method from another and it is this attribute, which can be amplified as is illustrated by the injection of high atomic number contrast agents or radionuclides, which target specific tissues. A new concept in imaging is the use of a pro-targeting agent such as a chemical which will seek the target of interest or be modified by the environment of interest such that when a signaling tracer or contrast agent is injected later that agent will attach only to the modified or sequestered pro-targeting agent. The intent of this presentation is to show the evolution of PET, SPECT and MRI from the perspectives of spatial resolution and contrast sensitivity. The limits of SPECT at 1 mm resolution for animal studies, of PET at 2 mm resolution for human studies and of MRI at 50 micrometers for the surface of the brain will be presented in the context of new scintillation detector materials, new imaging gantries and magnetic fields of 12 T. A presentation of an algorithm for Fresnel coded aperture imaging will initiate a discussion of other coded aperture approaches of interest to LLNL.

• Research Organization: Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
• Sponsoring Organization: USDOE
• DOE Contract Number: W-7405-ENG-48
• OSTI ID: 15013615
• Report Number(s): UCRL-PROC-200964; TRN: US0600195
• Resource Relation: Conference: Presented at: Signal and Imaging Sciences Workshop, Livermore, CA, United States, Nov 20 - Nov 21, 2003
• Country of Publication: United States
• Language: English

Related Subjects

42 ENGINEERING
62 RADIOLOGY AND NUCLEAR MEDICINE
70 PLASMA PHYSICS AND FUSION TECHNOLOGY
74 ATOMIC AND MOLECULAR PHYSICS
ATOMIC NUMBER
COMPUTERIZED TOMOGRAPHY
CROSS SECTIONS
DATA ACQUISITION
DATA TRANSMISSION
ELECTRON DENSITY
HELIUM IONS
LAWRENCE LIVERMORE NATIONAL LABORATORY
MAGNETIC FIELDS
MAGNETIC RESONANCE
NUCLEAR MEDICINE
PLASMA CONFINEMENT
RADIOISOTOPES
SCINTILLATION COUNTERS
SINGLE PHOTON EMISSION COMPUTED TOMOGRAPHY
SPATIAL DISTRIBUTION
SPATIAL RESOLUTION
TIME RESOLUTION