the size of the crystal. The X-ray photons then undergo photoelectric absorption Compton (or scatter subsequent with photoelectric absorption) resulting in the production of secondary electrons. The second function of the crystal is that of a phosphor-a solid which can transform the kine
tic energy of the secondary electrons into flashes of light. The geometrical design and the encapsulation of the crystal are such that most of these flashes of light leave the crystal through a side where they can be detected by a photomultiplier tube. Such scanners were fust developed by A.S.& E. Corporation(whicharenow sold by Pfizer Corporation). An example of a scanner of this type is also the EM1 7000 system which uses 1088 cesium iodide detectors and in each detector fan 1356 samples are taken. This system differs from the one depicted in Fig. 7 in one respect, namely the X-ray source rotates around the patient outside the detector ring. This makes it necessary to nutate the detector ring so that measurements figure may be made 1631. Fig. 8 likethoseshowninthe shows two reconstructions made such a scanner. on 3) Discussion o f Scan ConfigurationDifferences: An i m portant difference exists between the two scan configurations discussed here.The data in the fan-beam rotatingdetector scanners is essentially limited in the number of rays each pro-
Authorized licensed use limited to: Illinois Institute of Technology. Downloaded on January 30, 2010 at 11:33 from IEEE Xplore. Restrictions apply.
KAK: COMPUTERIZED TOMOGRAPHYa f i x e dr i n g ofdetectors
1253
anx-raysource r o t a t i n ga r o u n d t h ep a t i e n t
Fig. 7. A fixed detector-ring rotating source type scanner.
Fig. 8 . Two examples of reconstructions obtained on a fixed detector-ring rotating source type scanner. Note the superior resolution and contrast. (Courtesy of EMI.) (a) Noncontrast CT scan of the mid-brain demonstrating the third ventricle, frontal horns of the lateral ventricles, and quadrageminal cistern. Note the folia of the cerebellar vermis. (b) CT scan at the lumbarlevel. The psoas muscles are well demonstrated al as are the aorta and vena cava. Leaves of the mesentery are dearly seen. The w l of the colon is outlined by air and feces. Note the lumbar theca and anterior paraspinal veins. No contrast was used on the scan.
jection can have, although there is no limit on the number of projections. One can have only as many rays in each projection as the number of detectors in the detector array. On the
other hand, the data collected in the fixed detector scannersis limited in the number of projections that may be generated, while there is no limit on the number of rays in each projec-
Authorized licensed use limited to: Illinois Institute of Technology. Downloaded on January 30, 2010 at 11:33 from IEEE Xplore. Restrictions apply.
1254
PROCEEDINGS OF THE IEEE,1
VOL. 67, NO. 9, SEPTEMBER 1979I
tion.’ (It is now known that for good quality reconstructions the number of projections should be comparable to the number of rays in each projection.) In a fan-beam rotating detector scanner if one detector is defective, the same ray in every projection gets recorded incorrectly. Such correlatederrorsin all theprojectionsform ring artifacts[ 1081. On the othe
r hand when one detector fails in the fixed detector ring type scanners, it implies a loss or partial recording of one projection. When a large number of projections are measured, a loss of one projection usually does not noticeably degrade thequality of reconstruction[ 1091. The very nature of construction of the gas ionization detectors in the fan-beam rotational scanners lends them a certain degree of collimation whichis a protection againstreceiving scatter radiation. On theotherhand,thedetectorsinthe fixed ring scanners cannot be collimated since they must be capable of receiving photons from a large number of directions as the X-ray tube is rotating around the patient. This makes Fig. them more vulnerable to scatter radiation 861.[ B. Difficulties with the Interpretation of CTNumbers The reader will recall from the introduction to this section that the ray integral in (25) is given by ln(Ni,/Nd) only under the assumption that photons are monoenergetic. In practice this assumption is violated. Fig. 9 shows an example of an experimentally measured X-ray tube spectrum taken from Epp and Weiss[ 5 11 for an anode voltage of 105 kVp. For polychromatic photons (24)has to be replaced by6 (26) where&,(E) represents the incident photon number density (also called energy spectral density of the incident photons).&,(E) dE is the total number of incident photons in the energy range E and E+ dE. This equation incorporates the fact that the linear attenuation coefficient p at a point ( x, y ) is also a function of energy. The reader may note that if we were to measure the energy spectrum of exiting photons (on side B in Fig. 3) it would be given bySexit(E)=Sin(E)exP
Energy I n K e V
9. An example of an experimentally measuredX-ray trum. (From Epp and Weiss[ 5 1].)