Note: Descriptions are shown in the official language in which they were submitted.
CA 02283771 1999-09-27
IMAGE GENERATION METHOD
FIELD OF INVENTION
The present invention relates to an image generation method in a scintillation
camera, and
in particular to method for interpreting the electrical signals received from
an array of
photomultiplier tubes.
BACKGROUND OF THE INVENTION
In the human body, increased metabolic activity is associated with an increase
in emitted
radiation if the body is appropriately dosed with a radioactive tracer. In the
field of nuclear
medicine, increased metabolic activity within a patient is detected using a
radiation detector such
as a scintillation camera.
Scintillation cameras are well known in the art, and are used for medical
diagnostics. A
patient ingests, or inhales or is injected with a small quantity of a
radioactive isotope. The
radioactive isotope emits photons that are detected by a scintillation medium
in the scintillation
camera. The scintillation medium is commonly a sodium iodide crystal, BGO or
other. The
scintillation medium emits a small flash or scintillation of light, in
response to stimulating
radiation, such as from a patient. The intensity of the scintillation of light
is proportional to the
1
CA 02283771 1999-09-27
energy of the stimulating photon, such as a gamma photon. Note that the
relationship between
the intensity of the scintillation of light and the gamma photon is not
linear.
A conventional scintillation camera such as a gamma camera includes a detector
which
converts into electrical signals gamma rays emitted from a patient after
radioisotope has been
administered to the patient. The detector includes a scintillator and
photomultiplier tubes. The
gamma rays are directed to the scintillator which absorbs the radiation and
produces, in response,
a very small flash of light. An array of photodetectors, which are placed in
optical
communication with the scintillation crystal, converts these flashes into
electrical signals which
are subsequently processed. The processing enables the camera to produce an
image of the
distribution of the radioisotope within the patient.
Gamma radiation is emitted in all directions and it is necessary to collimate
the radiation
before the radiation impinges on the crystal scintillator. This is
accomplished by a collimator
which is a sheet of absorbing material, usually lead, perforated by relatively
narrow channels.
The collimator is detachably secured to the detector head, allowing the
collimator to be changed
to enable the detector head to be used with the different energies of isotope
to suit particular
characteristics of the patient study. A collimator may vary considerably in
weight to match the
isotope or study type.
2
CA 02283771 1999-09-27
Scintillation cameras are used to take five basic types of pictures: spot
views, whole body
views, partial whole body views, SPECT views, and whole body SPECT views.
A spot view is an image of a part of a patient. The area of the spot view is
less than or
equal to the size of the field of view of the gamma camera. In order to be
able to achieve a full
range of spot views, a gamma camera must be positionable at any location
relative to a patient.
One type of whole body view is a series of spot views fitted together such
that the whole
body of the patient may be viewed at one time. Another type of whole body view
is a continuous
scan of the whole body of the patient. A partial whole body view is simply a
whole body view
that covers only part of the body of the patient. In order to be able to
achieve a whole body view,
a gamma camera must be positionable at any location relative to a patient in
an automated
sequence of views.
The acronym "SPELT" stands for single photon emission computerized tomography.
A
SPELT view is a series of slice-like images of the patient. The slice-like
images are often, but
not necessarily, transversely oriented with respect to the patient. Each slice-
like image is made
up of multiple views taken at different angles around the patient, the data
from the various views
being combined to form the slice-like image. In order to be able to achieve a
SPELT view, a
scintillation camera must be rotatable around a patient, with the direction of
the detector head of
CA 02283771 1999-09-27
the scintillation camera pointing in a series of known and precise directions
such that reprojection
of the data can be accurately undertaken.
A whole body SPECT view is a series of parallel slice-like transverse images
of a patient.
Typically, a whole body SPECT view consists of sixty four spaced apart SPECT
views. A whole
body SPECT view results from the simultaneous generation of whole body and
SPECT image
data. In order to be able to achieve a whole body SPECT view, a scintillation
camera must be
rotatable around a patient, with the direction of the detector head of the
scintillation camera
pointing in a series of known and precise directions such that reprojection of
the data can be
accurately undertaken.
In generating an image with a nuclear scintillation camera, one of the
problems
encountered is that there is generally a shortage of detected gamma events.
One reason for the shortage of detected gamma events is that, for health
reasons, a patient
should be exposed to as little radiation as possible.
The image created by the scintillation camera is essentially a display of
detected gamma
events. If there are few counts, then there is little data to create the
image, and the image may be
meaningless from the point of view of human interpretation. It is not that the
resolution is poor;
it is just that the information is too sparse for a person to discern an
image.
4
CA 02283771 2003-05-O1
To generate an image from detected gamma events, the event information is
written into
an image or display matrix. Event by event, the data is written into picture
elements or pixels.
Each element or pixel contains input from zero to a high number of gamma
events, proportional
to the number of gamma events detected at the location corresponding to that
pixel. The more
gamma events, the brighter the pixel. A three dimensional graph of the pixels
can be generated,
showing the X and Y coordinates of the pixel locations in two dimensions, and
the number of
detected gamma events being indicated by the Z coordinate.
The collimator used in a scintillation camera provides the one to one spatial
correlation
of the emitted gamma rays at right angles to the crystal. The scintillation
crystal used in nuclear
scintillation cameras is sensitive. The collimator, however, reduces the
efficiency greatly as
gamma events occur in all directions, and as the collimator only lets through
the gamma events
that are substantially perpendicular to the scintillation crystal, most gamma
rays are absorbed by
the collimator. Collimators generally have efficiencies of minus four or five
orders of magnitude;
for example, for every 50,000 or so gamma events, only one passes through the
collimator and
is detected by the crystal.
Only a small amount of radioactive isotope can be administered to the patient,
and most
of the gamma events go undetected. With so few counts, an image will not have
enough
information to form a recognizable picture. As more counts are detected, a
pattern becomes
CA 02283771 1999-09-27
discernable; however, details of the pattern cannot be made out; for example,
the edge of an
object will not be discernable.
Since the patient can only be exposed to a limited amount of radioactivity,
one way to
generate a better image is to take the picture, i.e. detect emitted gamma
events, for a longer period
of time. However, there is a limit to the length of time for which a patient
can remain essentially
motionless. And in some cases, it is impossible for the patient to remain
motionless, such as
when it is the patient's heart that is being studied. It is common for studies
to last for about
twenty minutes, during which time the patient must attempt to remain as still
as possible as any
movement reduces the resolution of the generated image. As the study becomes
longer, it
becomes more difficult for a patient to remain still, and the resolution of
the image tends to
deteriorate.
One known method of dealing with the problem of a shortage of information is
to apply
a smoothing technique to the image data. Basically, smoothing techniques
involve moving a
certain amount of data from a pixel and moving it to surrounding pixels.
A typical technique or formula is a 121 242 121 smooth. The data associated
with a
particular pixel is assigned a weighting of 4 relative to its surrounding
pixels. The surrounding
orthogonal pixels are weighted as 2. The surrounding diagonal pixels are
weighted as 1.
6
CA 02283771 2003-05-O1
With respect to smoothing techniques, a heavy weighting means that the centre
pixel is
given an high weighting. An example would be a 121 2,20,2 121 smooth. A
relatively small
amount of data is assigned to surrounding pixels. This is referred to as a
light smooth.
With basic smoothing techniques as discussed above, the data is moved without
taking
into account characteristics of the data as a whole; i.e. the same smoothing
technique is applied
to each pixel, without taking into account information from other pixels. The
result is that the
edges of the image become blurred.
A more sophisticated smoothing technique involves weighting the centre pixel
by the
median value of the nine pixels in the immediate group. This is called a
median smooth. The
advantage is that one loses less resolution. The median smoothing technique
was developed for
looking at eye movements: since an eye generally looks quickly from one place
to another.
If certain pixels contain a large amount of information, the statistics are
good and little or
no smoothing may be needed. But in another area statistics may not be as good,
i.e. there may be
a shortage of data in the area of these pixels, and more smoothing may be
required. An
"intelligent" smoothing method, called a variable smooth, may then be used:
the more data there
is for a pixel, the more smoothing that will occur.
7
CA 02283771 2003-05-O1
Another smoothing technique examines the frequency content of the pixels.
Smoothing
is carried out in frequency space, or Fourier space. The resolution of the
system (i.e. the camera
that is writing the events into the pixels) can only resolve a certain spatial
frequency and not
higher. For example, with reference to the collimator, a camera may be able to
resolve 4 mm line
pairs (i.e. 2 mm of lead, 2mm gap). This will give a frequency of 4 line pairs
per cm. Any higher
frequency then cannot be resolved. In between is statistical noise that does
not really have a
meaning. Thus, the frequency content in the pixels is examined. If the
frequency content is
above what the system can resolve, then the excess frequencies are filtered
out.
Another smoothing technique uses a filter that implements a heavy smooth, and
subtracts
a light smooth and multiplied by a factor. Such a technique gives an edge
enhancement that
makes the image look better.
Smoothing techniques allow images to be discerned, but they do not add
information.
Such smoothing techniques simply spread out the known information so the
information can be
better interpreted by the human eye. However, in doing so, the spatial
resolution of the image
is compromised. In other words, the image looks better and patterns can be
seen, but, in terms of
information theory, information has actually been lost. It must be kept in
mind that one will
never be able to see something that cannot be seen from the raw or unsmoothed
data.
CA 02283771 2003-09-11
To review, smoothing is generally required to create a recognizable image from
insufficient data.. However, resolution is lost during the smoothing process.
SUMMARY OF THE INVENTION
An object of the invention is to provide an improved image generation method
in a
scintillation camera.
A second object of the invention is to provide an improved method for
interpreting the
electrical signals received from an array of photomultiplier tubes.
According to an aspect of the invention, there is provided a method of
generating
image corresponding to nuclear events. 'The nuclear events are obtained by a
nuclear
scintillation camera which has a scintillation crystal for detecting a
plurality of the nuclear
events and for generating a light scintillation corresponding to each detected
nuclear event,
and an array of photodetectors for detecting light scintillations generated by
the scintillation
crystal. Each of the photodetectors generates an electrical signal
corresponding to the
intensity of light detected by that photodetector. The method includes the
steps of: receiving
signals from a plurality of photodetectors; applying a plurality of
positioning algorithms to
the signals, each of which calculates a position of the nuclear event; and
generating image
data using a plurality of the positions obtained by a plurality of the
positioning algorithms.
According to a further aspect of the invention, there is provided a nuclear
scintillation
camera which includes: a scintillation crystal for detecting a plurality of
nuclear events and
CA 02283771 2003-09-11
for generating a light scintillation corresponding to each detected nuclear
event; an array of
photodetectors for detecting light scintillations generated by the
scintillation crystal, each
light scintillation being detected by a plurality of the photodetectors in the
array, each of the
photodetectors generating an electrical signal corresponding to the intensity
of light detected
by that photodetector; means for receiving the signals from the plurality of
photodetectors;
means for applying a plurality of positioning algorithms to the signals, each
of the positioning
algorithms calculating a position of the event; and means for generating image
data using a
plurality of the positions obtained by a plurality of the positioning
algorithms.
According to a further aspect of the invention, there is provided a method of
generating an image of the distribution and intensity of the nuclear events. A
nuclear
scintillation camera has a scintillation crystal for detecting a plurality of
nuclear events and
for generating a light scintillation corresponding, to each detected nuclear
event and an array
of photodetectors for detecting light scintillations generated by the
scintillation crystal. Each
light scintillation is detected by a plurality of the photodetectors in the
array. Each of the
plurality of photodetectors generates an electrical signal corresponding to
the intensity of
light detected by that photodetector. The method includes the steps of a)
receiving signals
from the plurality of photodetectors with respect to each nuclear event; b)
applying a plurality
of positioning algorithms to the signals to calculate a plurality of position
data, each position
data being generated by each respective positioning algorithm; and c)
producing an image
using the plurality of position data. When a small number of .nuclear events
are detected, a
recognizable image can be obtained.
CA 02283771 2003-09-11
According to a further aspect of the invention, there is provided a nuclear
scintillation
camera which includes: a) a scintillation crystal for detecting a plurality of
nuclear events and
for generating a light scintillation corresponding to each detected nuclear
event; b) an array of
photodetectors for detecting the light scintillation, each of the
photodetectors generating an
electrical signal corresponding to the intensity of light detected by that
photodetector; c)
means for receiving signals from the plurality of photodetectors with respect
to each nuclear
event; d) means for applying a plurality of positioning algorithms to the
signals to calculate a
plurality of position data, each position data being generated by each
respective positioning
algorithms; and e) means for producing an image using the plurality of
position data. When a
small number of nuclear events are detected, a recognizable image can be
obtained.
Other advantages, objects and features of the present invention will be
readily apparent to
those skilled in the art from a review of the following detailed description
of preferred
embodiments in conjunction with the accompanying drawings and claims.
to A
CA 02283771 2003-05-O1
BRIEF DESCRIPTION OF THE DRAWINGS
The embodiments of the invention will now be described with reference to the
accompanying drawings, in which:
Figure 1 is a sectional view taken along the line A-A of Figure 2 showing a
scintillation
camera head;
Figure 2A is a plan view of a scintillation camera head with thirty-seven
photomultiplier
tubes in a close packed hexagonal array.
Figure 2B is an enlarged view of the centre seven photomultiplier tubes with
regional
algorithmic application.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
Figure 1 shows the head (10) of a scintillation camera including thirty-seven
photomultiplier tubes or photomultipliers (11), a scintillation crystal (12),
a collimator (13), and
a housing structure ( 14) by which the components are held together in a
unitary manner. The
crystal (12) is a disc-shaped, planar scintillation crystal, such as thallium-
activated sodium iodide,
mounted in the housing (14) by means of suitable shoulders. Such crystals are
available in
different sizes; and a convenient size in wide use at present is 19 inches in
diameter.
11
CA 02283771 1999-09-27
The collimator (13) is interposed between the crystal (12) and the radiation
field (16) and
has a plurality of holes, the axes of which are perpendicular to the plane of
the crystal, for the
purpose of passing only those gamma rays which originate in the radiation
field in a region
directly beneath the hole. The photomultiplier tubes ( 11 ) are conventional
in nature and, with a
nineteen inch scintillation crystal, it is conventional to use thirty-seven
photomultipliers, each of
whose diameter is about 3 inches. The photomultipliers are perpendicular to
the plane of the
crystal, as shown in Figure 1, and the photocathodes of the photomultipliers
are spaced from the
upper surface of the crystal (12) in order to optimize the geometrical
sensitivity of the
photomultipliers. The spacing is chosen so that the geometric sensitivity is
constant and has the
largest value.
The gamma ray ( 17) emanating from a point in the radiation field ( 16) and
passing
through a hole in the collimator (13) above the point will enter the crystal
(12), and, depending
on its energy and the thickness of the crystal, will interact therewith at
some depth causing light
event (18) to occur. Such light event is seen by all photomultipliers. It is
the function of the
circuitry associated with the head (10) to compute the coordinates of the
point in the radiation
field causing the light event.
Before referring to this circuitry, it will be helpful to an understanding of
the invention
to define some general terms in connection with the array of photomultipliers
shown in Figure
12
CA 02283771 1999-09-27
2 since the invention is applicable to other arrays. In general, the motif of
any repeating pattern
of photomultipliers, regardless of their number, must include at least seven
photomultipliers.
Figure 1 shows the schematic of the system and its functions. The output
signals from
the photomultiplier tubes are connected to the preamplifier circuits and a
variable gain amplifier
controller(20) that are provided for each photomultiplier and are controlled
by the automatic gain
control system (22). The output signals from each preamplifier is connected to
the integrating
ADC (24) associated with each photomultiplier and the output of the ADC (24)
is transferred to
the energy selection circuit (42).
Undesirable events are rejected by the energy selection circuit (42) and the
desirable
events are transferred to the relative position circuit (44). The event
position from the energy
selection circuit (42) is calculated by the relative position circuit (44)
which passes the event
relative position information to the energy calculation circuit which loops
the information back
to the relative position circuit (44) to improve the precision of the
calculation. The output of the
energy calculation circuit (46) is used by the absolute position calculation
circuit (48) with the
data from the relative position circuit (44) to produce the absolute XY
location of the event to
increment the appropriate memory location of the computer memory (50) which is
subsequently
displayed by the display (52). It is to be understood that the above
description is not intended to
limit the scope of the invention as defined in the appended claims, as other
arrangements are
possible.
13
CA 02283771 1999-09-27
The position of the gamma event is determined by signals coming from multiple
photomultiplier tubes. It is known in the art to calculate the location of the
event using an
algorithm, and then to apply a smoothing technique. In accordance with the
present invention,
the location of such events are calculated using more than one algorithm.
Thus, by using more
than one method to calculate where the event occurred, i.e. use more than one
algorithm to
calculate the location of the event, then two positions are obtained from one
gamma event.
Where the gamma event occurs on the crystal is a single point. At that point a
light flash
or scintillation occurs. Since the light travels in many directions, the light
is generally detected
by more than one photomultiplier tube, and a number of photomultiplier tubes
generate signals.
The photomultiplier tube closest to the scintillation gets the most light and
has the strongest
output signal. These photomultiplier tube output signals can be used by more
than one algorithm
to calculate the position of the event. Note that when, for example, two
algorithms are used, the
number of events may have to be divided into two groups during certain
quantative analyses,
depending on their characteristics and relative efficiency of the two
algorithms.
If two algorithms are used, both generated images or image data can be written
into the
display matrix. The image statistics are thus improved because each algorithm
uses different
aspects and content of the data to derive the position of each event.
14
, ,
CA 02283771 1999-09-27
Two algorithms are used to calculate the location of each event prior to
assigning data to
one or more pixels. As long as the signals from the photomultiplier tubes are
used by the
algorithms in different ways, i.e. as long as the algorithms extract different
data from the
photomultiplier tube signals, then information is gained.
One of the ways that the processing algorithms should differ is in the way
that the
algorithms deal with the noise content of the signals from the photomultiplier
tubes.
A given algorithm may perform better with respect to signals received from
photomultiplier tubes corresponding to different areas of the scintillation
crystal. Certain
algorithms may perform better within regions of the scintillation crystal or
may perform better
overall throughout the crystal.
If more than one algorithm is used with a relatively low number of counts,
image quality
and perceived resolution is improved. This is the situation usually
encountered when using
scintillation cameras. If more than one algorithm is used with a relatively
high number of counts,
image quality and perceived resolution will generally deteriorate compared
with using one
algorithm because one of the algorithms will be superior. For a certain number
of counts image
quality and perceived resolution will be the same for both methods.
CA 02283771 1999-09-27
In one embodiment of the invention, two or more algorithms can be used until
good
statistics are obtained. After a time, for the pixels with good information,
just the best algorithm
may be used; the data obtained by the second best algorithm may either be used
or discarded.
The present invention enables better images to be obtained in the same length
of time.
Similarly, the invention enables similar images to be obtained in less time.
While obtaining
superior visual quality, less accuracy is lost compared with smoothing
techniques.
The resolution of the camera system RS is given by the formula:
R _ WRQ+WRb 2+R2
s
R.a is the resolution of the first algorithm; Re is the resolution of the
second algorithm; It~
is the resolution of the collimator; RS is the resolution of the system, Wa
and Wb are weighting
factors depending on event fractions and correlations of the algorithms.
A gamma ray passing through collimator plate encounters the scintillation
crystal which
produces light. The crystal does not always produce exactly the same amount of
light. The
photomultiplier tubes convert light to an electrical signal, which is not
always exactly the same.
Adjacent photomultiplier tubes may generate the following signals, for
example: 90/800/110,
100/805/100, 110/795/95. This sets the limit of the intrinsic resolution, i.e.
the resolution of the
16
CA 02283771 1999-09-27
crystal and photomultiplier tube assembly. The system resolution is the square
root of the sum
of the squares of the intrinsic resolution and the collimator resolution. The
collimator resolution
relates to the range of angles at which gamma rays can pass through the
collimator, which
depends on the apertures in the collimator.
By using a second algorithm, for example, the system resolution may change
from 7.74
mm (with a single algorithm) to 7.78 mm (with two algorithms, given an Ra=3.3;
Re=3.5; R~=7).
For this relatively small reduction in resolution twice the number of counts
are generated and the
image is improved.
Examples of algorithms that can be used may be referred to as the centroid
algorithms and
the circle algorithm. However, this invention is not algorithm specific, and a
number of suitable
algorithms can be selected by one skilled in the art.
It should also be noted that the algorithms are to be weighted. One algorithm
might be
superior in one area, another in a second area, and yet another in a third
area. Three algorithms
could be differently weighted depending on where the light event occurs
relative to the
photomultiplier tubes.
17
CA 02283771 2003-05-O1
With the reference to Figure 1 the scintillation camera system comprises the
digital
camera, energy rejection circuit 42, relative position calculation 44, energy
calculation circuit
46, absolute position calculation circuit 48.
In its preferred embodiment, the energy rejection calculation is digital and
independent
of the relative position calculation, which means that it can be performed,
before, or in parallel
with the relative position calculation. If it is performed after relative
position calculation then it
becomes position dependent. If the energy correction is performed before the
relative position
calculation, events which are outside the required energy window can be
filtered earlier in the
process, which improves the efficiency, and hence the speed of the
positioning.
In the preferred embodiment it is assumed that a tuning device exists, as
described in
United States Patent No. 5,237, 173 but not limited to such devices, and that
the tuning is done
before the acquisition for the energy information and positional information.
The assumption is
that before acquisition, tuning is performed on the detector head, which will
normalize the
responses of all the light detectors. The assumption is that the detector head
is digital, but not
limited to being digital. (This energy correction method can be used with any
detector head on
the market, which can improve the characteristics of the detector heads.)
After or instead of those
tuning devices, a new calibration is also performed based on a hole phantom
image acquisition.
18
CA 02283771 2003-05-O1
Outputs from the digital detector head as seen in Figure 1, are the following:
1. The label or sequential number associated with the light detector in the
detector
head T, with the highest response, or in the close neighbourhood of the
detector with the highest
response. The light detector with the highest response or in close
neighbourhood will be called
the centre light detector. The assumption is that the absolute coordinates of
each light detector is
known in the detector head.
2. The response signal of the central light detector of an n-tuple, defining
the n-tuple
as a group of the light detectors in the neighbourhood of the centre light
detector.
3. The responses of all light detectors in the neighbouring n-tuple of the
central light
detector, defining the n-tuple as a group of the light detectors in the
neighbourhood of the centre
light detector.
Energy rejection circuit 42 produces a sum signal of said n-tuple of light
detector signals
including the signal of the central light detector. Relative Position
calculation circuit 44
produces x and y values for the particular n-tuple of the light detectors.
Output from the position
calculation is the associated label or sequential number T of the centre light
detector in the n-
tuple.
19
CA 02283771 1999-09-27
Energy rejection circuit 42 let pass the events with an energy within the peak
energy
window. For those events, relative position calculation and energy calculation
are weakly
dependent. Energy calculation may give back an energy evaluation to the
relative position
calculation, which improves the precision of the position. This loop may be
done zero, one or
more times.
The energy calculation method consists of three well defined phases: first,
acquisition of
the energy information; second producing the energy calculation tables; third,
applying the energy
calculation 46 in real time acquisitions.
Acquisition of energy information: For each of many n-tuples with
corresponding central
light detector in the preferred embodiment, N by M histograms are recorded
which cover the area
of calculation of one n-tuple. Each histogram consists of at least 256 bins.
Histograms are
addressed by the highest n bits of the x position and the highest m bits of
the y position. For each
event with particular position x and y, particular histogram is chosen
depending on position, and
the counter of that histogram is increased, depending on the energy. The
number of counts in
each histogram has to be statistically sufficient. Acquisition is done with
the known energy, and
without any structured phantoms or collimators.
For producing the energy tables; in the preferred embodiment, histograms
should be
filtered with a 3D filter for each n-tuple to smooth the response. It is known
in the prior art that
CA 02283771 1999-09-27
the response of the light detectors is higher in the centre, and it decreases
towards the periphery
of the light detector, and that the response is continuous. Responses of the n-
tuples are also
smooth. For each n-tuple, the maximum response of each of the histograms is
computed after
filtering. For each histogram the factor should be computed so that the
responses of all the light
detectors are equal. For each n-tuple, a table of N by M factors is stored in
the energy table.
When applying the energy calculation 46 in real time, for each event, and
depending on
the central light detector of the n-tuple, address or label, and also
depending on the first m bits
of x coordinate and n bits of y coordinate, a particular address in the table
is addressed. The
computed energy, which is the sum of all the signals in the n-tuple of light
detectors including
the central light detector, is multiplied by the factor in the table. This
produces the energy
calculated value for that event.
In the preferred embodiment, the relative position calculation method consists
of four well
defined phases. First, acquisition of the position information; second,
producing the position
calculation tables for each light detector in the n-tuple and third applying
the relative position
calculation 46 in real time acquisitions. The fourth phase consists of adding
the relative position
of the n-tuple to the known geometric position of that n-tuple in the
scintillation detector to create
the absolute position 48. Assumption is that the detector head is capable of
providing:
21
CA 02283771 1999-09-27
1. Associated label of the light detector in the detector head, with the
highest
response, or in the close neighbourhood. We will call the light detector with
the highest response
in one event the centre light detector.
2. Assumption is that the absolute coordinate of each light detector is known
in the
detector head.
3. Responses of all the light detectors in the neighbouring n-tuple, defining
the n-
tuple as a group of the light detectors, in the vicinity of the centre light
detector.
4. In preferred embodiment n-tuple is consisting of seven or more light
detectors.
5. Definition of the event: Event is one incidence of the gamma photon
producing
the scintillation effect in the crystal of the detector head. Detector head
outputs the label T of the
centre light detector, and the values of the centre light detector and the
intensity values of the light
detectors in the neighbouring n-tuple.
6. Positional calculation is the translation of the events from the light
detectors output
to X, Y and energy values.
22
CA 02283771 2003-05-O1
In the acquisition of position information; acquisition consists of two parts.
First,
acquisition with the structured phantom in front of the scintillation camera
(similar to Smith
phantom), and second, acquisition without phantom, the so-called flood
acquisition. Smith
phantom is known in the art, and consists of a lead plate with lots of
pinholes in a rectangular
array. The preferred embodiment uses a hexagonal pattern of holes array, with
cycle harmonized
to the disposition of the light detectors within the detector head. A
mechanism is added to the
hexagonal lead plate such that, by manoeuvring one of three levers, the plate
may be moved half
a distance between two neighbouring holes, so that the resolution along the
three axes defining
the hexagonal pattern is doubled. Acquisition is done with the radioactive
isotope having a
known energy. For each of many n-tuples with a corresponding central light
detector, in the
preferred embodiment, image data is acquired. The images are distorted
depending on the
geometric arrangement or constellation of the light detectors, the light
detector and electronic
channel properties, and the method of the position calculation. The position
of each pinhole from
the phantom is determined. The second acquisition of the flood is needed to
determine that the
uniformity criterion is satisfied. The uniformity criterion is related to the
number of counts in
each area in between the position determined by the image of the pinholes and
bounded by the
splines which connect all the positions of the pinholes in horizontal and
vertical direction. The
number of counts in each image has to be statistically sufficient to determine
the position of the
pinholes, or to check if the uniformity criterion is satisfied.
23
CA 02283771 2003-05-O1
To apply the relative position calculation 44 in real time; for each event,
and depending
on the central light detector of the n-tuple, address or label, and also
depending on each light
detector signal of the n-tuple, a particular address in the table is
addressed, which gives a distance
from the scintillation to the light detector centre. This is done for each
light detector, giving a n-
tuple of said distances. Position calculation is performed by solving the
linear system of
distances. This produces the position calculated value for that event.
Circuit 48 calculates the absolute position correction in real time. For each
event, after
calculation of the relative addresses and depending on the central light
detector of the n-tuple,
address or label, the position of the n-tuple is added to the relative
position inside the n-tuple to
form the absolute address.
In the preferred embodiment, the position calculation method consist of three
well defined
phases. First, acquisition of the position information, with one radioactive
isotope with lower
energy (approximately 100 keV) and later with the radioactive isotopes in the
medium (250 keV)
and high energy ranges (511 keV). Second, producing the expansion correction
factors in table
form or function with interpolation for the energies between the acquired
energies.
In the preferred embodiment, to improve the energy independent position
correction
method consists of three well defined phases. First, acquisition of the
position information; with
one radioactive isotope with lower energy (approximately 100 keV), and later
with the radioactive
24
CA 02283771 2003-05-O1
isotopes in the medium (250 keV) and high energy ranges (511 keV). Second,
producing the
expansion correction factors in table form or function with interpolation for
the energies between
the acquired energies. In circuit 46, the expansion correction factors are
applied to the X, Y
values calculated in 44, together with the sum of the light detectors values (
E given by the energy
rejection circuit 42). Although the preferred embodiment illustrates a purely
digital camera, it is
to be understood that the above described methods can be easily adapted to
operate when analog
position calculation is used.
The centroid algorithm is used to calculate the incident location of a gamma
ray on the
crystal of a gamma ray camera detector head. The primary inputs to the
algorithm are the energy
response signals of an array of photomultiplier tubes (PMTS) that lie on the
opposite side of the
detector head crystal. PMTs are small (3" diameter) cylindrical devices that
detect light rays and
output an electrical signal proportional to the intensity of the detected
light. The further an event
occurred from a PMT the lower its signal.
The first step of the centroid algorithm is to calculate the centroid, or the
intensity
weighted averaged position of the event. This is calculated by summing the
product of the
position of each PMT and its energy response to an event, and dividing this
sum by the sum of the
intensities of each PMT. This can be expressed mathematically as
0 0 0
cX = apmtX * pmte / apmte and cy = apmty * pmte / ~pmte
CA 02283771 2003-05-O1
where (cX, cy) is the centroid calculated position, (pmtx, pmts) is the
position of a PMT, pmte is
the energy response of this PMT, and the sum is over all the PMTS. This first
calculation is very
approximate, and weights all events toward the centre of the PMT under which
the event
happened. As such, it is not clinically useful without further corrections.
The first such correction is that for linearity. A linearity correction table
is created by
exposing a known configuration of gamma emitting point sources and calculating
the first step
centroid for each gamma event. This results in an image of the point sources
whose locations have
been moved from their actual position due to the known tendency of the
centroid method to skew
events toward the centre of each PMT. The linearity correction table contains
the correlations
between the known ("real") position of the gamma emitting source, and the
position calculated
by the centroid equation. Applying the linearity correction table to the
centroid image of the
point sources will produce an image of the point sources in their "real"
orientation.
Uniformity and energy corrections are subsequently applied to the image, but
these are
not particular to the centroid algorithm. Linearity correction is also not
exclusive to the centroid
method, but is absolutely mandatory. Analog cameras use linearity corrections
as well, but the
pre-linearity corrected images from analog cameras are much closer to the real
image than with
digital cameras employing the centroid method.
26
CA 02283771 1999-09-27
Another algorithm which may be used to calculate the incident location of a
gamma ray
on the crystal of the gamma camera detector is the "circles" algorithm. As in
the centroid
algorithm the inputs to the calculation are the measured energy responses of
the photomultiplier
tubes to the incident gamma ray.
The principle of this algorithm is that the energy response of the
photomultiplier tube to
a fixed energy incident gamma ray is non-linearly, but monotonically related
to the distance of
the point of incidence from the centre of that tube via the so called "roll-
off' curve, and
consequently the energy may be used to determine a circle of possible
incidence around each
tube. The radius is given as a function of energy r = R(E). Taking the radii
and centres of the
circles for several such tubes responding to a given gamma event, allows the
calculation of a
common point of intersection of the circles, localising the point of incidence
absolutely.
The incident point (X,Y) of the gamma ray may be calculated after, suitable
approximations, using the equation:
X = 1 ~Z ~ rzk xk
with a similar form for y. The sum is over the photomultipliers responding to
the event, Z is a
normalisation constant, rk is the radius from the centre of the kt'' PMT, and
xk is the x position
of the PMT.
27
CA 02283771 1999-09-27
As for the centroid algorithm, the resulting point is approximate and requires
corrections
for non-linearities of the system.
Numerous modifications, variations and adaptations may be made to the
particular
embodiments of the invention described above without departing from the scope
of the invention,
which is defined in the claims.
28
