3 4
versing of the entire subject by transmission photons greatly within an energy window 14, the event is stored in the
increases the opportunity for scatter. uncorrected image memory 16. The memory 16 stores the
number of events or the count at each x,y location. The
SUMMARY OF THE INVENTION window 14 typically represents the photopeak area of the
The method for eliminating scattered photons from trans- 5 detected enerSy spectrum. For example, the photopeak winmission or emission images includes collecting a planar dow ^ correspond to plus or minus 10 percent about a photon image formed by collecting a count of detected Une of ^ radioisotope in use. For Tc-99m, a photons in a large matrix of pixels, each pixel including sPectni ^ exists at 140 keV and the corresponding phoposition coordinates and an energy spectrum for photons *°peak area is 126 keV to 154 keV. collected at the position coordinates. The image is divided 10 The data fr°m the camera 12 also follows a path where it into a plurality of local areas, each local area including at is stored in a multichannel acquisition memory 18. The least one pixel. A local energy centroid is calculated for each memory 18 stores not only the position (at the detector) of local area and a total energy centroid for the image is the event but also its energy, thus each position or pixel has calculated. a corresponding energy spectrum consisting of the number
The spectra of local areas having respective energy cen- 15 of events or counts at various energy levels. Data from the
troids less than a known threshold percentage of the total memory 18 is analyzed to create a correction table 20 The
energy centroid are used to estimate a local scatter correction values fr°mthe correction table 20 are combined with the
factor for each local area that are applied to the image. dataln *e memo^ 16 tou CTeat^ 3 corrected image 22. The
corrected image 22 may be used directly, or more typically,
BRIEF DESCRIPTION OF THE DRAWINGS 20 undergoes further processing such as additional error correction and/or inclusion into 3-D images.
FIG. 1 is a schematic diagram of a nuclear camera for Referrmg to FIGS. 3 and 4, a flow chart shows the
producing emission images. anal is rformed on the ^ stored m me muitichannel
FIG. 2 is a schematic diagram of a nuclear camera for memory 18 to create the correction table 20 which is applied
producing transmission images. 25 t0 the uncorrected image in memory 16 to produce the
FIG. 3 is a block diagram of a data processing system corrected image 22.
suitable for performing a method according to the invention. Sufficient counts need to have occurred at the positions of
FIG. 4 is a flow chart diagram of a method according to interest in an image to produce statistically reliable results,
the invention. 3Q For example, 30 counts provides for reasonably reliable
FIG. 5 is an axial end view of a cylindrical phantom. data.
FIG. 6 is a partial plan view of the phantom of FIG. 5 as 111 some cases> mere ^ not be enough positions or
it is to be imaged. pixels with sufficient counts to proceed on a pixel by pixel
TM~, _ . , . .c. . basis. In this case pixels can be grouped together to provide
FIG. 7 is an exemplary graph of the camera image along sufficient data the line S—5 of FIG. 6 showing exemplary data with and 35
without scatter correction according to the invention. For sample, pixels may be grouped together in the form
o • ., i j- r . , • of 2 pixel by 2 pixel (4 pixel by 4 pixel, 8 pixel by 8 pixel,
naSis a blockdiagram ofammtichannelacquisi^ and so on) local areas or boxes.
memory suitable for use with the invention. „, , ,
Choosing local areas larger than a single pixel also may
DESCRIPTION OF THE PREFERRED 40 be desirable for reducing computational overhead, but at the
EMBODIMENT expense of spatial resolution.
Referring to FIG. 1, a simple nuclear camera or gamma * me case of £local ff8 mf a sfSle P^1' an camera for producing emission images consists of a planar ener§y ... for each local area is formed by adding the image detector 2. A subject 3 contains a radioisotope the sPectra of the "^dividual pixels within the local area, emits photons 3 that are detected by the detector 2 to form 45 Whether a local area consists of lxl pixel or greater, an an image of the activity within the subject 2. energy centroid is calculated for each local area based on the
Referring to FIG. 2, a simple nuclear camera for produc- counts obtained at various energy levels. The energy cening transmission images consists of a planar image detector tiolds m calculated by the first momenta method. 2 and a photon source 4 (a radioisotope). Photons 3' travel A*1 of me local area spectra are added and a total energy through the subject 1' to the detector 2. An image of the centroid is calculated for the summed spectra, subject 1'is formed at the detector 2 from the attenuation and It has been found desirable also to exclude local areas other effects upon the transmission through the subject 1' of having very few counts from the process (e.g., less than 30 photons from the source 4. counts).
These cameras can be combined by adding the source 4 to 55 It has been discovered that the best measures of scatter the camera of FIG. 1. This allows transmission studies to be come from the local areas having a local energy centroid less performed to measure such parameters as attenuation by the than a threshold percentage of the total energy centroid. This subject. These parameters can then be used to improve the percentage may be, for example, 95 to 99 percent, with 99 quality of emission studies. percent providing excellent results.
Referring to FIG. 3, a data processing system 10 receives 60 ^ a local energy centroid for a local area is less than the event data from a gamma camera 12. For each event, this threshold, the corresponding spectrum for that local area is data includes the x-coordinate and y-coordinate of the event included in a sum of all such spectra to produce a global in the detection plane, along with the energy value. The scatter curve.
system 10 may be implemented in specialized hardware or The next task is to locally normalize the global scatter as part of a general purpose computer. 65 curve to provide a local scatter curve for each local area.
The data from the camera 12 branches in one path to a It has been discovered that a small window just below the conventional image memory 16. If the energy of an event is low side of the photopeak is the optimal-area for determin
