USER'S GUIDE TO 3-D PET SIMULATION SYSTEM
Introduction
PET images are reconstructed from measured projection data of
positron source distributions. This simulation system can generate
realistic projection data from computer-simulated objects, normally
the segmented MRI data. Projection data are reconstructed by the
filtered backprojection algorithm to obtain simulated PET images.
Image protocols
scan_type = 1 MR, FDG or CBF imaging, fuzzy phantom
scan_type = 2 FDG or CBF imaging, clear phantom
scan_type = 3 Transmission imaging, clear phantom
scan_type = 4 F-DOPA Imaging, clear phantom
scan_type = 5 Transmission imaging, fuzzy phantom
scan_type = 6 F-DOPA with caudate shrinkage
scan_type = 7 blank scan with a rod source
scan_type = 8 transmission scan with a rod source
Program name
petsimu
Synopsis
petsimu <studyname> <scantype> [<options>] > ! logfile &
Description
petsimu is a script to control the PET simulation system. It sends
info about the parameters/progress of the simulation to stdout.
This is usually redirected to a logfile with a patient specific name
for record. Its I/O file structures are as follows:
Input
infile ! 3-D organ model images
tacfile ! object creation parameter
parfile ! projection simulation parameter
recfile ! image reconstruction paramter
Output
studyname.att ! attenuation lengths (cm)
studyname.att_hd ! header file
studyname.bla ! blank projection data (counts)
studyname.bla_hd ! header file
studyname.tra ! transmission data (counts)
studyname.tra_hd ! header file
studyname.dat ! emission data (counts)
studyname.dat_hd ! header file
studyname_FBP_itm.mnc ! transmission image (1/cm)
studyname_FBP_ima.mnc ! emission image (nCi/cc)
studyname_t_e1.mnc ! axial weighted image
studyname_tr_e1.mnc ! true tra image
studyname_e_e1.mnc ! axial weighted image
studyname_pet_e1.mnc ! true emi image
studyname_voi_e1.mnc ! resampled tissue maps
Infile is usually the segmented MRI data registered with real PET
images. It is in the form of studyname_string.mnc with a patient
or subject name and any user-specified string. A typical filename
is miche_seg_e1.mnc.
MRI model contains gray matter, white matter, CSF,ventricle,tumor,
skull bone and a set of functional anatomical structures. They are
obtained using mincclassify/msp and roi programs at the BIC with
the following coding convention:
GM WM CSF VEN MS BRN BKG BON LCN RCN LPU RPU
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0 1 2 3 4 5 6 8 21 22 23 24
Each code identifies a tissue type or structure. MS and BRN are
for internal tissue types other than GM, WM, CSF and VEN. BON
refers to skull bone and BKG everything else. LCN, RCN, LPU
and RPU are for left and right caudate nucleus and putamen in
the basel ganglia. Of course they can be used to represent any
structures as long as user knows what they are.
Table 1: Tissue attenuation coefficients for 511 Kev photons
Tissues Electron Density (1/cm3) Att (1/cm) Ratio
--------------------------------------------------------------
Air 3.622E+20 0.00010 0.001
Blood 3.513E+23 0.10066 1.051
Bone 5.267E+23 0.15092 1.575
Brain 3.438E+23 0.09851 1.028
BrainStem 3.500E+23 0.10029 1.047
Cerebellum 3.428E+23 0.09823 1.025
Cerebrum 3.433E+23 0.09837 1.027
CSF 3.420E+23 0.09800 1.023
Fat 3.061E+23 0.08771 0.916
Hair 4.202E+23 0.12040 1.257
Heart 3.415E+23 0.09785 1.021
Lucite 3.833E+23 0.10983 1.146
Lung 8.607E+22 0.02466 0.257
Muscle 3.445E+23 0.09871 1.030
Polystyrene 3.380E+23 0.09685 1.011
Skin 3.639E+23 0.10427 1.088
Water 3.343E+23 0.09579 1.000
--------------------------------------------------------------
Usage
User begins a simulation session by editing I/O information in the
script petsimu. Copy parameter files tacfile, parfile, recfile from
/usr/people/may/demo and edit them according to the header info
therein and specific applications.
User should first run att scan to get attenuation data for emission
simulation. This step is done once for each study. Later emission
protocols for the same subject can use the same att data to model
photon attenuation effect.