Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
10~478
"Device for measuring the distribution of the absorp-
tion or the emission of radiation in a layer of a
body"
The invention relates to a device for
determining the spatial distribution of the absorp-
tion or the emission of radiation in a layer of a
body, the absorption or the emission of the body being
measured in a large number of measuring series in a
large number of directions in the layer by means of
a radiation source and a detector, each measuring
series producing a number of measuring values of the
absorption or the emission in strips which extend at
least approximately parallel relative to each other,
the absorption or the emission being calculated and
displayed in image points of the layer on the basis
of the said measuring values. This device is prefer-
ably used for X-ray diagnosis or nuclear medicine.
A device of this kind is known from
United States Patent 3,778,612, Hounsfield, December 11,
1973. The absorption in a (human) body is measured by
means of a radiator which is displaced, together with a
radiation detector which measures the radiation behind
the body, perpendicularly to the direction of the
radiation, the detector measuring a series of measuring
values ~measuring series) which is a measure for the ab-
sorption of the radiation along the straight lines
through the body which extend parallel to each other and
which are determined by the position of the radiator
and the detector. After such a measuring series, the
radiator/detector system is rotated and a further
measuring series is completed at a different angle
- 2 -
j PHD.75-o64
~ ~0~1478 23-4-1976
relative to the body~ etc. The absorption in the
individual points or regions in the layer covered
by the measurement cannot be simply reconstructed
from thc measuring values obtained, because the
measuring values do not represent a measure for the
absorption in individual points, but rather for the
absorption in a straight line or strip through the
body to be examines which is-obtained during the
¦ measurement. Erom an arithmetical point of view, this
implies that the value of a function (absorption, emis-
~ sion, density, etc.) in individual points of the
¦ layer defined by the straight lines must be calculated
from the traject integrals of this function along
a large number o~ intersecting straight lines.
This problem is also encountered in
~the measurement of the radioactivity distribution
in radioactively marked biological objects, in the
calculation of layers of macromolecules (viruses and
the like) measured by means of an electron micros-
cope, and also in the examination of layers of
technical objects (for example, rnaterials test) by
- means of penetrating radiation. The reconstruction
of the absorption in the layer is effected in known
~ ~QP)i/~
devices in that the layer to be ex~m-ilff~-is sub-
divided into a matrix of square image elements, the
dimension of which corresponds approximately to the
width of a strip. Each image element is assigned the
measuring value (or a value derived therefrom and
from the other measuring values of the measuring
series) of each measuring series which has been
measured in the strip in which the image element is
situated. ~f it is assumed that the image element and
,
1~;147~
the strip have approximately the same width, an image element can be
influenced by as much as three measuring values (or measuring values
derived therefrom) which represent the absorption of the rad~ation in
three parallel strips. The measuring value or the value for a strip
derived therefrom, conse~uently, is multiplied by a weighting factor
for the calculation of an image element, the said factor corresponding
to the common plane of the strip and the image element.
This method, performed in a computer, involves very long
computing times and a very expensive computer, notably for the calculation
of the weighting factors.
In order to realize a shorter computing time by means of a
simple device, it has already been proposed to superpose the values
derived from the measuring values by a convolution process on the target
of a charge storage tube along adjoining strips, the position and the
direction of the said strips corresponding to the position and the
direction of the strips during the formation of the assigned measuring
values. However, serious problems are then encountered as regards the
signal-to-noise ratio.
The invention has for its object to realize fast processing of
the measuring values by simple means without deterioration of the signal-
to-noise ratio.
According to this invention there is provided a device for
determining the spatial distribution of the absorp-tion or the emission
of radiation in a layer of a body, the absorption or the emission of the
body being measured in a large number of measuring series in a large
number of directions in the layer by means of a radiation source and a
detector, each measuring series producing a number of measuring values of
the absorption or the emission in strips which extend at least approximately
parallel relative to each other, the absorption or the emission being
calculated and displayed in image points of the layer on the basis of
the said measuring values characterized in that there is provided an
allocation device which assigns the measuring value, or a value derived
4 -
, .
10~1478
therefrom, to each image point of the layer to be examined for each
measuring series, the said
- 4a -
,~.
.
PIID.75-064
1061478 23-l~-1976
valuc hnving been obtained in a strip in which the
relevant image point is situated, the values assigned
to the image points from all measuring series being
stored and superposed in a storage and summing device,
the superposed values being displayed by means of a
display device, each time the mean value of the
absorption or the emission in different adjacent
image points in the region determined by these image
points being displayed.
The invention will be described in
detail hereinafter with reference to the drawing.
Figure 1 shows the geometrical lay-out
of the strips and the layer to be examined,
Figure 2 shows a first embodiment in
accordance with the invention~
Figure 3 shows the geometrical lay-out
of the strips, the layer to be examined and the image
points for the embodiment shown in Figure 4,
Figure 4 shows a further embodiment, and
Figure 5 shows the appearance in the
time of the clock pulses in the device shown in Figure 4.
The absorption of the object 0 in the
layer E to be examined is determined by way of a large
number of meas~ring series, the absorption being
measured each time along a number of parallel ex-
tending strips. Figure 1 shows these strips 1...6
for a measuring series. The absorption in each of
these strips is represented by a measuring value.
This measuring value is subsequently subjected to a
convolution process, the value M1.. M6 thus calculated
including, besides the measuring value measured in
the strip, the weighted sum of all other measuring
P~ID.75-OG4
1061478 23-4-1976
values of this measuring series.
The calculation of the absorption from
these values M1...M6 is effected in that the values
- are "spread" across the region of the layer to be
examined, the said region being covered by the asso-
ciated strip; for example, the value M2 is assigned
to the part of the layer which is covered by the
strip 2. Consequently, the position of the point
determines which value is assigned to a point of the
layer E. Subsequently, the values derived from a
further measuring series are assigned, the said values
being superposed on the previously assigned values.
The assignment of a point of the layer
to a strip is effected by means of a computer which
calculates, on the basis of the position of a point
in a fixed x, y system of co-ordinates, its relation-
ship with one of the strips. The association of a
point of the layer with a given strip is dependent
of the distance 7 (see Figure 1) between a straight
line which passes through the point, parallel to the
strip, and the co-ordinate origin (x = 0, y = 0),
the said distance being calculated from the formula:
~ = x . cos e + y.sin e (1)
Therein, x and y represent the position of this point
in the x, y system of co-ordinates, whilst e is the
angle at which the strips intersect the x-axis. The
distance ~ is calculated for each individual point
of the layer, it then being possible to calculate
directly the strip whose value (for examp1e, M5) is
to be assigned to the point (the relationship be-
tween the distance ~ and the strlp whose value M
is to be assigned to points at the distance ~ is
Pl]~.75-oG4
23-l~-1976
1061478
clearly illustrated by Flgure 1).
Figure 2 shows a device in accordance
with the invention which is constructed on the basis
of the foregoing considerations, The device comprises
two sawtooth generators 10 and 11 which generate two
sawtooth signals UX and uy of different frequency.
If it is assumed that UX and u are proportional to
the distance x and y, respectively, from the co-ordi-
nate origin, the output signals UX and u of the
sawtooth generators 10 and 11 represent a quantity
which line-wise scans the layer to be examined (Figure
1 shows only two lines ~ and 9).
It is important that the distance
between two adjacent lines is substantially smaller,
for example, by a factor 3 or 4, than the width of
a strip. In the configuration shown in Figure 1, in
which the width of the layer corresponds to four
strip widths, this means that the layer must be scan-
ned by approximately 12 to 16 lines. This implies
that the period of the signal u must be 12 to 16
times larger than the period of the signal ux. In
practice, the layer is not scanned in six rather wide
strips, as is shown in Figure 1, but along a large
number of strips, for example 150 strips. The number
f lines must then be accordingly larger (450 to 600)
so that the horizontal and verticai deflection gene-
rations (the latter generators scanning 625 lines)
commonly used in a television apparatus are suitable
for use as the sawtooth generators 10 and 11.
The output signals UX and uy of the
sawtooth generators 10 and 11, respectively, are
each applied to the input of a multiplier circuit 12
--7--
PI I D, 7 5 ~
10~1478 23-4-1 97G
and 13, respectively, the other input of said multi-
plier circuits carrying a voltage which is proportion~l
to cos e and sin e, respectively. The output signals
of the multiplier circuits 12 and 13 are added in an
adding circuit 14 in which an additional value ~ 0
is added so that, taking into account the equation
(1), the output of the adding circuit 14 carries a
signal which is proportional to ~ + ~ . This signal
is applied to an analog-to-digital converter 15 which
converts the output signal into a digital number. When
the proportions are suitably chosen, the digital
output signal represents the number of the strip or
the address of the store in which the measuring value
obtained in this strip or the value M derived there-
from is stored. This will be illustrated on the
basis of the below calculating example.
It is assumed that the distance y or
x = 1 fromthe co-ordinate origin corresponds to the
width of the strip. It is also assumed that the value
x or y = 1 corresponds to the signal UX or uy = 1 Y,
and that no further proportionality factors are
introduced by the multiplier circuits 12 and 13 and
the adding circuit 14, so that, for example, for
e = o, x = 0.5, a voltage of 0.5 V is present on the
output of the adding circuit 14, the value correspon-
ding to ~ 0, assumed to be 4 V, being added thereto
yet. If a voltage of, for example~4.5 V has thus been
generated, it will be converted into the numerical
term 4.5 by the analog-to-digital converter 15, how-
ever, this converter 15 is constructed so that the
last decimal position is eliminated, so that on the
output of the converter 15 the number 4 is present
--8--
PI-II) 75-o64
106~478 23-4-1976
which, if the values M1...M6 are correctly assigned
to the storage position, denotes the address of the
storage position in which the value M4 assigned to the
strip 4 is stored. Because the sawtooth voltage UX
increases, the output voltage always increases and
_ exceeds the voltage value 5V at a given instant, for
example, when for e = 0 the value x becomes larger
than 1. The number 5 then appears on the output of
the analog-to-digital converter 15, which means that
the address of the store in which the measuring value
M5 assigned to the fifth strip is stored is addressed.
Generally, the analog signal is converted into a
binary value for which the above considerations are
also valid.
The addressing device formed by the
analog-to-digital converter 15 controls an inter-
mediate store 16, the various storage positions of
this store storing the values measured in the indi-
vidual strips or the values (M1...M6) derived there-
from, so that each time the contents of the addressed
storage position are available on the output of the
intermediate store 16.
The intermediate store 16 can be re-
placed by controlled multiplex access to data present
in analog form. The value each time.fetched from the
intermediate store 16 in this manner is written on a
disc store 17 which simultaneously synchronizes the
sawtooth generators 10 and 11. The measuring values,
or the values M1.,.M6 of a single measuring series
derived therefrom, assigned to the individual points,
are written on a single track in this disc store 17.
For the next measuring series, the
PIID 75-o64
23-4-1976
10~1478
strips then intersecting the layer to be examined at
a different angle e, for example, a process computer
introduces new values sin e and cos e (for this reason,
the multiplier circuits 12 and 13 are preferably con-
structed as multiplying analog-to-digital converters),
and the measuring values, or the values derived there-
from, which are obtained in this series and which
usually deviate from the measuring values, or the
values M1...M6 derived therefrom of the preceding
series are written in the intermediate store 16. These
new measuring values, or the values derived therefrom,
are assigned to the layer one point after the other,
and are stored accordingly in the next track of the
disc store 17.
This is repeated for all measuring
series, so that the number of tracks of the disc
store 17 should correspond at least to the number
of measuring series recorded.
When all measuring series have been
processed and recorded on different tracks in the
disc store 17 in this manner, all tracks are simul-
taneously read during a read operation. The signals
read are added in an adding circuit 18 and are applied
to a display apparatus 19, for example, a television
monitor. This display apparatus 19 has a limited
resolution so as to achieve the blurring of the
various lines to form an image element. The limit
of the resolution could possibly be imposed by a strip
width. This means that the width of the line whereby
the layer is scanned is smaller, for example, by
the factor 3 to 4, than the resolution of the display
apparatus 19.
--1 O~
1'111~. 75_ofil1
23~1~_1976
1061478
Thus, far, it has been assumed that
the absorption is measured in exactly parallel strips
within a measuring series. However, there are also
devices for measuring the absorption in a layer of
a body in which a large number of detectors cover the
flared radiation of the radiator behind the object.
In these devices, the strips in which the absorption
is measured diverge, viewed from the radiator. In
~` such a case, the equation (1) must be replaced by
1Q ~ (x,~) = 1-+~S(x s~iYn-~in y cos e) (2)
Therein, ~ describes the divergence of the radiation
beam and amounts to zero in the border case of paral-
lel projection. In the case of such flared spreading
of the measuring values or the values derived there-
from, the adding circuit 14 must be replaced by an
analog computing network which reproduces the formula
(2).
In the described embodiment in accor-
dance with the invention, each point of the layer to
be examined is continuously assigned the value
(M1...M6) measured along the strip in which the re-
levant image point is situated. The formation of the
mean value of the absorption in different adjacent
image points and the display of this mean value in the
region determined by these image points is also
continuously effected, in that the-;mage, superposed
by means of the adding circuit 18 which adds the
video signals of the individual tracks, is displayed
on a display device 19 having a limited resolution.
Figures 3, 4 and 5 show a further em-
bodiment in accordance with the invention.
In order to assign the strips obtained
~061478 PHD 75-o64
by the "spreading" of the measuring values to an image
matrix representing the layer to be examined,(Figure
3 is based on an image matrix of only 4 x 4 4uadratic
elements~ each element, whose dimensions correspond
approximately to the width of a strip, is again sub-
- divided into a matrix of points as shown in Figure 3
which illustrates this sub-division for one element.
These points are assigned, without interpolation, to
the strip in which they are situated. As a result,
the exact movement is replaced by an approximative
movement as is indicated in Figure 3 by the heavy,
stepped line 20. Generally, a sub-division of an
element into 3 x 3 points already suffice.s. The number
of image points whereto a measuring value or a value
derived therefrom is to be assigned is thus increased
by a factor 9, but each point is assigned each time
to only one strip. Calculation of any weighting fac-
tors is thus avoided.
Figure 4 shows a device for reali~ing
a fast reconstruction of the absorption values in
the layer to be examined; Figure 5 shows the asso-
ciated clock signals. The device mainly consists of
two units, one of which effects the assignment of a
measuring value or a value derived therefrom to a
point of the layer, whilst the othe~ unit calculates
on the basis of the values assign~ to the individual .
points of an element, the mean value of the absorption
for this element and, after this has been done for
all measuring series, this unit superposes and dis-
plays the mean values assigned to this image element.
The assignment of an image point to a
strip, i.e. to a measuring value or to a value derived
-12~
rllD.75-ef~
1061478 23-4-1'~76
thererrom, is defined by the cquation (1) when parallel
projections are assumed. In the case of a flared
projection, the distance is calculated in accordance
with the equation (2). If the strip width ~ ~ = 1 is
chosen by appropriate standardization, the result of
the equation (1) merely need be rounded off to an
integer number in order to produce the number (address)
of the intermediate store position in which the
measuring value (or the value derived therefrom)
measured in the strip at this distance is stored. If
the constant value 0.5 is added to the numerical
value of the equation (1), the rounding off in the
upward or downward direction is replaced simply by
elimination of the positions behind the decimal
point. Figure 4 can be described on the basis of
this consideration.
The addends ~ x cos e and ~ y sin ~
are prepared by two shift registers 21 and 22 whose
contents are cyclically advanced per image by a clock
signal, ~ x representing the distance between two
points in the x-direction, and ~ y representing the
distance between two adjacent points in the y-direc-Jc
tion. However, these addends can also be calculated
by a process computer. This is particularly advan-
tageous when the angles at which the radiation passes
through the layer to be examined during the measuring
series have not been predetermined.
When all points are sequentially pro-
cessed line-wise, the value ~ for each new point must
be increased by the addend ~ x cos e and for each new
line by the addend ~ y sin e, starting from initial
-13-
1061478 23_i~_1976
values xO, yO which correspond to the co~ordinates
of the first point of a line or of an image and which
have been written in the mono-cell interrnediate stores
23 and 24' by the clock signals ty and tB. This is
effected by means of three adding circuits 24, 25
and 26. In the adding circuit 24, in conjunction
with the store 23 whose output is fed back to an
input of the adding circuit 24, the clock pulse tx
generates a (preferably digital) signal whose instan-
taneous value varies in accordance with the co-ordi-
nates of thepoints on a line. This step-wise increas-
ing signal is returned to the initial value xO by
the clock signal ty after each line. In the adding
circuit 25, the same clock signal generates, in con-
junction with theintermediate store 24', a signal
which increases in phase and which corresponds to
the y-co-ordinate of the relevant line (y perpendicu-
lar to the line direction). Each time when all image
points of the layer have thus been treated, the
store 24' is returned to its initial value yO by the
signal tB-
The initial values xO and yO are depen-
dent of the angle e; however, an initial value can
always be chosen at random, for example xO = 0, it
then being necessary to change the other ~alue accor-
dingly. This other initial value can be given in
advance by the device which also supplies the addends
~ x cos e and ~ y sin e, i.e. by a shift register
22' which comprises a number of storage cells which
corresponds to the number of measuring series and
whose contents are cyclically circulated by the
clock pulse tB.
-14-
rl ll) , 7 5 - o G 1~
10~1478 23-ll-1976
The output signal of the intermediate
stores 23 and 24' is applied to both inputs of an
adding circuit 26, the output signal of which controls
an addressing device 27. In the case of a digital out-
put signal of the adder - obtained in the described
- manner - the least significant bits of a digital
signal can be omitted if correct standardization and
generation of the address are chosen. If the signal
which represents the distance ~ is present in analog
form, the addressing device must comprise an analog-
to-digital converter as has already been described
with reference to Figure 2.
The device to be described below could
in principle be constructed as shown in Figure 2.
The embodiment shown in Figure 4, however, digitally
processes the values M1...M6, so that the construc-
tion is different.
The address formed by the addressing
device 27 is used for directly actuating the random
access intermediate store 28 in which the values
M1...M6 are stored and whose output supplies the
desired value for further processing.
The subsequent device sums the values
assigned to the individual points of an element,
superposes the measur~ng values obt~ined from the
individual measuring series, and s~ores the recon-
structed image. The values assigned to three adjacent
points are each time added in a loop between an adding
device 29, having an input which is connected to the
output of the intermediate store 28, an intermediate
loop 30 comprising a single storage position, and a
switch 31, under the control of clock signals t1 and t2
l'llD.75-o61~
1061478 23-4-197~
(see Figures 4 and 5). l~hen t~le first value is taken
over, provided that the switch 31 is in the correct
position, previous values are also added, i.e, the
value assigned to a set of three points o~ the same
element in one or two previous lines, or the value
assigned to an element on the basis of preceding
measuring series. The value added is dependent of
the position of a second switch 32 which is connected
to a contact of the switch 31 and whose other contact
is connected to the output of a shift register 33
which is connected behind the intermediate store 30
which may also be considered as a shift register ~com-
prising one storage position).
When the value of a line of points has
thus been obtained and taken over, under the control
of the clock signal t2, in the subsequent shift
register 33 which comprises m-l-storage positions
(m = number of elements of a line; so 4 in the present
example) three successive lines are added by means of
the clock signals t2 and t3 in that the value of the
previous line is fed back to the adding device 29
via the switches 31 and 32 The combination of the
3 x 3 points to form one image element is thus per-
formed without an additional storage position being
used. -~
The complete line of image elements is
stored in a shift register 34 (number of storage posi-
tions: n-m; n = number of elements of the image matrix)
by means of the clock signal t3. A-t the beginning of
each line of elements, that is at the beginning of
each third line of points, a stored value is returned
from the shift register 34 to the adder 29, via the
-16-
Pllr). 7s-oG
23~ 1976
10~1478
switches 31 and 32 and under the control of the clock
signals t2, t3 and t3 in order to superpose the ab-
sorption distribution obtained from the new measuring
series on the absorption distributions obtained from
previous measuring series. After completion of the
superposition, the absorption distribution derived
from the relevant measuring series is stored in the
network formed by the shift registers 30, 33 and 34
(possibly after superposition of the previously ob-
tained absorption distributions) and can be read.
Before the beginning of a new reconstruction,that
is to say when completely new measuring series have
to be evaluated, obviously, the stores 30, 33 and 34
must be set to zero. This is not separately shown
in the circuit diagram.
Figure 5 shows the variation in time of
the clock signals tx, ty, tB, t2 and t3. The clock
signals tB~ t2 and t3 correspond to the signals tB~
t2 and t3, respectively; however, the individual pulses
are wider, i.e. they start earlier and terminate later
than in the clock signals without a stroke. The
clock signal t1 corresponds to the clock signal tx,
but has been delayedwith respect thereto, because
the reading of a value Ml...M6 from a line of the
intermediate store 28 cannot be effected simultaneously
with the supply of the address of this storage
position.
The references I-1 and I-2 in the time
diagram of ~igure 5 denote two successive periods of
the reconstruction of two complete images; L-1, L-2..... L-4
denote the periods of the reconstruction of a line
of the image I-1 or I-2, and P denotes the reconstruc-
17
PIID,75-~)GII
1061478 23~ 976
tiOIl period of an image element.
Referring to the device shown in Figure
4 and the diagram shown in Figure 5,it is to be noted
that the operation is only diagrammatically illustrated.
In order to ensure a simple operation of the stores,
normal steps known in the digital technique must be
taken. For example, a shift register comprising
charge-coupled elements must be actuated by two clock
pulse series, supplied to different inputs. The
feeding back of a store output to the preceding
adding device (like, for example, the components 29
and 30) ~equires an additional intermediate store
(not shown in the drawing) in order to prevent, in
the case of a modification of an output quantity,
the unambiguous state of the adding device from being
disturbed be~ore it is taken over in the store. These
problems, however, occur in all logic circuits, so
that it not necessary to elaborate this aspect.
The device shown has the advantage
that the sub-division of an element into different
points is controlled only by clock signals. For exam-
ple, if an ~lement is to be sub-divided into a dif-
ferent number of points, for example, 4 x 4 points
instead of 3 x 3 points, merely the clock drive need
be modified.
-18-
. .
