CA1073121A - Tomography - Google Patents

Abstract

ABSTRACT OF THE DISCLOSURE
For examining a body by means of X-rays or other penetrating radiation, data is determined representing the absorption suffered by each of a plurality of beans which are passed through the body. The beams are effectively arranged in a plurality of sets of beams, in a single plane, at many different angles. The arrangement is such that the amount of information provided by the beams is a function, which is in part non-uniform, of the beam position in the set. A reconstruction of the distribution of absorption of the radiation within a part of the body is derived from the absorption data so determined.

Description

~73~

This invention relates to a method of and apparatus for examining a body by means of radiation such as X or y radiation.
The method and apparatus according to the invention can be used to assist in the production of radio~raphs in any convenient form, such as a picture on a cathode ray tube or other image forming device r a photograph of such a picture, or a map of absorption coefficients such as may be produced by a digital computer and on which contours may subsequently be drawn.
In the method of, and apparatus, for examining a body described and claimed in British Patent Specification No.
1,283,915 radiation is directed through part of the body, from an external source, in the form of a pencil beam. A scanning movement is imposed on the beam so that it takes up in turn a large number of differing dispositions, and a detector is used to provide a measure of the absorption of the beam in each such dispostion after the beam has passed through the body. So that the beam takes up these various dispositions the source and the detector are reciprocated in a plane and are orbited about an axis normal to the plane. The various dispositions thus lie in a plane through the body over which the distribution of absorp-tion coefficient, for the radiation used, is derived by proces-sing the beam absorption data provided by the detector. The ; processing is such that the finally displayed distribution of absorption i~ the result of successive approximations.
The method and apparatus described in the aforesaid British patent has proved to be successful for producing cross-sectional representation of parts of the living body, such as ~ ;
the head.

In my co-pending Canadian Patent Application Serial No. 198,145 filed 16th April 1974 there is descrlbed a further method and apparatus having a method of data acqu:isition : .
.,. ~ . ~ - ~. ;, . . .
. ' ' ' ' . ' , .! , 3~2~L
essentially the same as that referred to in regard to the afore-said British Patent Specification while the method of processing of the data is more flexible and differs for the reason that it is based upon a convolution technic~ue.
One advantage of employing a convolution technic~ue to derive an image of the absorption distribution in the exploring plane is that, unlike the iterative method of reconstruction described in the aforesa.id British Patent Specification, it is not necessary to reconstruct the ~hole of the absorption pattern in the exploring plane in order simply to reconstruct a part, rather if a special locality alone is of interest this region only may be made the subject of reconstruction, with economy, for instance, in time of reconstruction. The ability to recon-struct the absorption pattern over a limited area of interest is of particular value in the examination of parts of a body of large cross-sectional area as in -the example of the human torso.
It is undesir~ble however on grounds of economy of equipment, given that the area over which it is required to ex-: am.ine closely will not normally amount to more than a minor fraction of the total cross-sectional area, for the apparatus to operate with the ability to resolve the pattern over the total area in fine detail. Fol.lowing out the technique described in said Canadian Application, however, the apparatus would be sub-ject to this objection.
The arrangement described herein shows how it is pos-sible to overcome this difficulty.
According to the invention there is provicled apparatus, for examining a body b~ means of penetrating radiation, such as X-radiation, comprising a source of a fan-shaped distribution of said.radiation arranged to project said radiation along a plura-lity of paths through a slice of the body, means for scanning said source relative to said body to project said rad:iation through said slice along further paths, detector means for 1~73~2~
providing output signa~sindicative of the amount of absorption suffered by said radiation on traversing said paths, said detec-tor means including a pluralit~ of detectors, adjacent ones of which are disposed to receive radiation along respective paths ~ inclined to each other at a given angle, and means being provi-; ded for sampling said detectors in groups at interleaved times to derive therefrom output signals relating to sets of substan-tially parallel paths, neighbouring sets being inclined to each other at an angle greater than said given angle.
In order that the invention may be clearly understood and readily carried into effect one example thereof will now be described with reference to the accompanying drawings of which:-Figure 1 shows the general layout in side elevation of an apparatus in accordance with the invention; ~;
Figure 2 illustrates the same apparatus in end eleva-tion;
Figures 3 and 4 illustrate parts of the apparatus in relation to the use of a liquid medium surrounding the patient's body in the immediate vicinity of the region under examination;
Figures 5 and 6 illustrate features of the detecting means used to detect the exploring radiation ater its passage through the patient's body;
Figure 7 is an explanatory diagram relating to the dis-tribution of the multiplicity of radiation sensing devices used in the detecting means as described with reference to Figures 5 and 6;
Figure 8 shows in diagrammatic form the general layout of the entire apparatus including those parts concerned with pro-cessing of the absorption data;
Figure 9 illustrates a feature of the apparatus, and Figure 10 illustrates means or deriving absorption data in a form suitable in particular to the special processing.
In Figure 1 a patient 1 is shown lying on supporting ~L~73::~Z~ -means formed in two parts, 2 and 3 and his body is subject to examination by X-radiation indicated in broken line at ~. This radiation is generated by a source 5 and forms a fan shaped spread in a plane lying at right angles to the plane of the paper. It will be appreciated that the patient supporting means r has to be sufficiently long to allow any desired sectior of the patient's body to be located in the plane of the X-radiation.
In the region of the exploring radiation, the body of the patient is surrounded by a liquid medium, which may be water, 10 and which has an absorption coafficient for the radiation close-ly-similar to that of body tissue. The liquid is shown in the figure at 6 and contained within an envelope, or bag 7. The envelope 7 is positioned within a ring like structure 8 which may be of metal such as duralumin.
The ring member 8 is held by retaining means not shown in the figure, and an important featuxe of this means is that it allows traverse of the ring member 8, together with the patient, along the direction of the axis of the ring, and moreover allows of displacement of this member in the plane of the exploring 20 radiation in any direction. Thus a particular cross section of the body of the patient can be selected for examination by lon-gitudinal traverse of the ring member 8 and the patient. The displacement possible in a direction normal to the axis of tra-~erse permits of a local area of the cross section selected to be examined in fine detail as will be explained more fully later.
With displacement of the ring member 7 at right angles to the axis of longitudinal traverse, the parts 2 and 3 of the patient supporting means are arranged by suitable means to under-go similar displacement, and a support 9 for the part 2 is arran-30 ged to allow of this though the means is not shown in the figure.
The part 3 is supported at its end remote from the ring member 8 by one or more rollers 10. Each roller 10 is carried on a bea-ring supported by an axle member 11, which member has an axis - . . .:. :: ~ . . . , ; . . .

~I~!i7312~L

about which the orbiting motion of the X-ray source 5 takes place as will be made more clear. The support of the part 3 by the roller 10 allows of the displacement of the part 3 along with the ring member 8 when this is displaced laterally for the purpose of local area selection. At the other end of the part 3 from the roller 10 the part 3 is hinged at 12 to the ring mem-~, ber retaining means, thus allowing of vertical displacement of the member 8 for the purpose of local area selection.
Around the body of the patient when he is located in position in the apparatus there is disposed a surround or frame13 which is cylindrical along its length having a longitudinal axis which is the axis of the axle member 11. At its end adja-cent this latter member it is closed and supported by a bearing 14 which in turn is supported by the member 11. At its other end it is open to allow of positioning of the patient within it, and at this end it is supported on rollers 15 which have suitable fixed bearings. These rollers 15 are such that the surround member 13 is free to rotate on its axis, which as has been indi-cated is the axis about which the orbiting motion of the X-ray source 5 takes place. The source 5 is mounted on the surround member 13 by means of a support 16. Directly opposite the source 5 there is mounted on the surround member 13, by means of a sup-port 17, a deteator means 18 so as to provide radiation absorp-tion data from the body of the patient in the plane of the radi-ation from the source 5.
The axle member 11 is carried by a support 19 and adja-cent the support 19 and surrounding the axle member 11 is a bob-bin 20. This last element is fixed to the support 19 and wound round it are cables 21 and 11, respectively carrying absorption data from the detector means 18 to the procecising unit and sup-plying power for the X-ray source 5. With the oxbiting motion of the source and detector means the cables wind on or off the bobbin 20. They are fed to the bobbin via guides 23 and 24 3~
; respectively which are carried by the surround member 13. This ; member may make one or more orbiting revolutions and the cables wrap or unwrap in relation to the bobbin 20 correspondingly.
At the bobbin the cables are secured and thence pass to their respective connecting units, namely the data processing unit mentioned, and a power supply unit.
Figure 2 as stated shows an end view of the apparatus illustrated in Figure 2 and elements 5, 8, 13, 15, 16, 17 and -~ 18 have the same significance as in Figure 1. At 30 in Figure ` 10 2 there is indicated the location of the orbiting axis and 31 shows the outline of the cross section of the patient's body in the plane of the exploring radiation. The circle 32 lying with-in this cross section, and centred upon the orbiting axis 30, defines a selected local area, namely the area contained within it, over which the processin~ unit which processes the absorp-tion data derived from the detector means 18 operates to provide high resolution information concerning the absorption distribu-tion of the patient's body in the examined cross section~ The selection of the local area, as has been indicated earlier, is accomplished by appropria.te displacement of the patient's body in a direction normal to the orbiting axis of the apparatus, the displacement illustrated in Figure 2 being primarily a lateral one.
Figure 2 furthermore shows particular rays 33, 34, 35 and 36 emanating from the radiation source 5. Rays 33 and 34 lie tangentially with respect to the circle 32 enclosing the selected local area, and rays 35 and 36 lie on the extreme edges of the fan of radiation from the source 5. As will be explained more fully the radiation lying between the limits set by the rays 33 and 34 is subdivided into narrow beams to provide absorption data while outside these limits the radiation is subdivided into broader beams. As will be seen rom the figure, the detector means 18 extends over the whole spread of the fan of radiation 1~73~L2~
from the source S r namely from ray 35 at one extreme of the fan to ray 36 at the other extreme. . .
Referring to Figure 3, the ring member 8, and liquid medium 6, for positioning the patient in the apparatus is again shown in relation to the surround member 13, but in rather more .
: detail than in Figure 1. Thus as shown .in Figure 3 the member 8 is flanged at its ends as indicated in the figure at 40 to :::
increase its rigidity, and split at 41 into two halves, namely . a lower half 81, and an upper half 82, those halves being rela-tively located by suitable means such as pins, for example, not : shown in the figure. The liquid medium 6, which as stated ear- ~
lier may be water, is contained within a wrap-round form of ~:
envelope, or bag, 42, corresponding to 7 in Figure 1. This bag is located by the cylindrical portion of the ring member 8 ly-ing intermediate its flanged ends. Contained within the bag and ring member the patient's body is constrained to occupy some .
. displaced position within the surround member 13 as required by . .:
the selection of the local area for examination in special de-tail. :~
In Figure 4 the arrangement is shown with the upper half 82 of the ring member 8 removed, and the bag 42 lying un-wrapped over the lower half 81 of the ring member, this half being disposed in undisplaced relation with respect to the sur-round member 13. The arrangement shown is such as might be the case immediately prior to the entry of the patient into the ; apparatus. With entry, the bag 42 is wrapped round the patient in the region of required examination, the upper half of ring 8 is fitted into place and secured in position, and the bag is inflated with the liquid medium so that the medium fills all the ,~
space between the patient's body and the ring. The patient and ring together are then moved axially of the surround member 13 until the examination region is brought under the X-ray source 5, and patient and ring are then displaced normally with respect to --7 i .

. , . ., .. ~ . ...... . . .

312~
.
the axis of 13, namely the orbital axis of the apparatus, for the required local area selection. A number of ring members such as 8, but of differing diameters, may be used, that member fitting most closely around the patient being chosen, so that minimum absorption of X-ray photons occurs in the liquid medium 6.
It will be realized that particularly with extreme displacement of the examined cross section in a direc-tion away from the orbital axis of the apparatus there will be a tendency ; 10 for certain rays of the fan of radiation, to be subject to large variations of overall absorption in the course of the orbital motion of the apparatus. Absorbing means, such as shaped blocks of the materiaI known by the registered trade mark "Perspex"
and indicated in Figure 8 by reference numeral 69 are preferably provided to mitigate this effect. Other variants of the said apparatus are also described in the said copending patent appli-cation.
Figure 5 shows the arrangement of detectors in the detector means 18 referred tv in relation to Figures 1 and 2.
This means has the object of providing the absorption data which on suitable processing, such as will be described subsequently~
enables an image to be reconstructed of the cross section, of the patient's body, examined by means of the exploring radiation from a point source. In Figure 5 the point X denotes the point source of the radia*ion, this point source orbiting about the a~is, at O, of the apparatus. The figure shows in broken line the extreme circular bound 13' centred on point O, if the pa-tient's body in any possible position. The circle 32 represents the bound of the area for which image reconstruction is effected with high resolution. The circle 32 is also centred on ~ and any area of the cross-section of the patient's bocly which it is desired to reconstruct in high resolution must necessarily be located within the area contained by 32.

;

3~ Z:~L
In diagrammatic manner various rays are shown pro-ceding from tha point source X and these rays pa~s through the ; area within the bound 13' to fall on a multiplicity of radia~
tion sensitive devices denoted in the figure by 43 and 44. It will be seen that in so far as the rays from X pass through ; the region bounded by the circle 32 they are shown as rela-tively many and closely spaced, whereas those rays lying more towards the extremes of the fan and not passing through the circle 32 are shown as comparatively few and widely spaced. In this respect the figure illustrates diagrammatically the prin-ciple mentioned earlier that the selected area of the cross-section of the patient's body concerning which information is ;~
re~uired in fine detail is examined by closely spaced narrow beams whereas areas lying outside the selected area are explored by relatively broad widely spaced beams. It will be realiz~d that to the extent that the radiation sensitive devices 43 and 44 collect photons of the radiation they each correspondingly define a beam of radiation.
Collimators, not shown, are located in front of the radiation sensitive devices to define the apertures of thedevices and respective heams. The radiation sensitive devices denoted by 43 in the figure have apertures oE relatively small width but are closely packecl. These define many beams passing through the selected area defined by the circle 32. The remaining radiation sensitive devices 44 have apertures of relatively larger width and define broader beams. The widths of the various beams defined in the way just described will be referred to in greater detail hereinafter.
The outer beams may also be of considerably reduced intensity with the corresponding and added advantage of reduc~
tion of the dose of X ray, to the patient. By this means, and by reconstructing the absorption pattern in fine detail only over a limited area, the reduction in dose, as colmpared with _g_ :

' 1~73i~23L

reconstruction of the whole area of cross-section in such detail, may be in the ratio of 4 : 1.
The radiation sensitive devices 43 and 44 take the form of so-called scintillation crystals and each crystal when ~; irradiated from the source 5 generates light which is incident upon an associated photo-multiplier. The respective photo-multipliers are not shown in Figure 5 in the interests of sim-plicity, but they have the function of trans~orming the light output from the respective crystals into electric currents which are fed to the processing equipment for the purpose of image reconstruction. The scintillation crystals may be of ;
sodium iodide type, such as is commonly used for scintillation purposes.
The photo-multipliers associated respectively with the scintillation crystals of the detector means 18 are rela-tively bulky and they present the problem of accommodating them ~; conveniently in the apparatus.
Figure 6 shows a suitable way in which the photo-multipliers may be disposed.
In this figure, which presupposes the radiation source to lie to the right, the rays lQl, 102, 103 .. are to be re-garded as representative of the relatively narrow beams of Figure 5 falling on scintillation crystals such as 43. The ray 101 may be considered as representing an extreme beam of this set of beams. The location 111 marked on it is to be taken as the location of the scintillation crystal on which the beam is incident. Centred on the location 111 there is shown the photo-multiplier 111' which is excited by the scintillation of the crystal at 111. The photo-multiplier 111' is shown drawn in full line and this is intended to signify that the photo-multiplier lies to one particular side of the plane of the exploring beams of radiation~ The adjacent beam represen-ted by the ray 102 falls upon a corresponding scintillation 1~73~
; ~ .
-; crystal located at 112 and excites a photo-multiplier 112'.
This photo-multiplier is shown drawn in bro~en line to indi-cate that it lies on the other side of the plane of the ex ploring beams to photo-multiplier 111'. The beam represented by ray 103 falls on a scin-tillation crystal at 113 to excite photo-multiplier 113'. This photomultiplier lies on the same side of the exploring beams as 111'. Continuing, the beam represented by ray 104 falls on a scintillation crystal loca-ted at 114 with excitation of photomultiplier 114'. This photomultiplier is disposed on the side o-E the beams remote from photomultiplier 111' and 113'. The pattern of this arran-gement proceeds similarly for rays 105, 106, 107, 108, but with ~;
ray 109 scintillation crystal is located after the same manner as is the scintillation crystal in the case of ray 101. From this point/ the cycle of disposition of the photomultipliers repeats, and continues repeating until all the rays represen-ting the relatively narrow beams are accounted for.
Figure 7 illustrates the distribution of relatively narrow and relatively broad beams across the fan of radiation emitted by the source 5. A point to be noted is that the vari-ous beams are relatively divergent, but as will be explained later, the data is assembled in sets relating to the absorption suffered by parallel beams and the data processing proceeds on the basis of parallel sets. In Figure 7, and for this reason, the arrangement of beams is shown as if they were in fact parallel. This being effectively so, the figure illustrates the passage of a set of parallel beams through the re~ion con-tained within the perimeter 13' wi-thin which the cross- sec~
tion of the patient must lie.
As in Figure 5, the point O denotes the location of the axis of orbital rotation and 32 the circle within which image reconstruction of a selected area of the cross section of the patient's body is to be reconstructed in fin~ detail.

~L0~3~L2~l Concentric with 32 and lying within 32~s the circle 45, and within this circle image reconstruction conforms to a parti-cular degree of accuracy regardless of whatever absorbing material may be present outside of the boundary 32.
In the figure, 46 designates a boundary which is tangential to the circle 45, and 46' similarly designates a boundary diametrically opposite and tangential also to the circle 45. Between the boundaries 46 and 46' there are a total ;~
of 80 parallel exploring beams each having a mean width of lmm in this example of the invention. The boundary 47 parallel to the boundary 46 is tangential to the circle 32 on the same side of the point O as boundary 4~. In like manner boundary `~
47' parallel to boundary 46' is tangential to circle 32 in diametrically opposite fashion to boundary 47. Between the boundaries 46 and 47 and between rays 46' and 47' there are in each case a total of, in this example, 13 parallel beams each of mean width lmm. Boundary 48 is drawn parallel with boundary 47 on the same side of the orbital axis O and on the other side of the axis ~he boundary 48' is drawn in similar relation to the boundary 47. Between each of these pairs of boundaries there is one single beam of 3mm mean width. Further-more, boundary 49 is parallel with boundary 48 on the same side of the axis at 0, while boundary 49' on the other side of the axis is disposed in identically similar relation to 48'. Be-tween each of these pairs of boundaries there is one beam of ~ lOmm mean width. Finally, th~ extreme boundary 50, parallel ; to the boundary 49, just touches the circle 13l on the same side of the axis at O while on the other side the boundary 50' : is likewise situated in relation to boundary 49'. Between these last two pairs of boundaries there is in each case one beam o mean width 55mm. It will b0 understood that in refer-ring to the beams described in relation to Figure 7 as parallel beams, or in referred to any set of parallel beams, the ~L073:1Z~ ~
parallelism is to be understood as the paralleIism of one beam to another rather than that each beam in itself is strictly a parallel beam. The references to the beam width in the fore~
going is the width as determined by the collimators measured along a line perpendicular to a central ray passing through the point 0. The centre lines of adjacent narrow beams in the central area are moreover 2mm apart, ancl the gaps between them are filled in with other beams as will k,e explained later. In fact the effective beam width is wider than lmm, because of spreading caused by effective presence of a "scanning aperture".
It will be appreciated that other distributions of narrow and broad beams may be utilized. Furthermore each of the broad beams may be replaced by a single narrow beam. In that case the absorption measured by such a narrow beam would be used as the absorption value for each of a number of narrow beam dispositions which would otherwise have covered the broad beam region. Such an arrangement would also give the reduction in-X-ray intensity referred to above.
Figure 8 sets out diagrammatically the general layout ZO of an entire apparatus of which the scanning part is illustra-ted in Figures 1 to 4.
In this figure, the point X again denotes the point of emission of X-radiation from the source 5, the point O the location of the orbital axis, the circle 32 the area of high .. ~, resolution, 13' the area within which the cross--s~ction con-cerned must be located, 18 the detector means providing absorp-tion data for processing and 69 the absorbing means referred to hereinbefore but not shown in the earlier figures.
The block 51 represents a store and auxiliary com-ponents for recei~ing and holding absorption data from the detector means 18 as it is produced in the course of the orbi-tal motion of the apparatus. The block 51 also includes res-pective amplifiers 56 for the output currents from the various 73~
photomultipliers of -the detector means 18 as they are received in the unit 51. The gains o~ the amplifiers are individually adjusted to compensate for the differing sensitivities of the various scintillation crystals of the detector means 18~ The various output currents from the amplifiers are respectively integrated by Miller integrator circuits 57, the outputs of these circuits being respectively converted from analogue to digital form by converters 58 before storage occurs. If de-sired the gains of the amplifiers may be commonly controlled to compensate for any variations that may occur in the emission intensity of the X-ray source.
It is desired that the final image reconstruction shall represent the distribution of the absorption coefficient over the area of the cross sectional material under examination.
That absorption coefficient is the absorption per unit length in the immediate vicinity of a given point of an exploring beam passing the point. To achieve the required result, it is necessary that each output signal derived from the detector means 18 shall be converted to its logarithmic form. For this purpose the unit 51 includes a log-converter 59 comprising logarithmic look-up tablesaccording to known usage. Each sig-nal from the integrator, after conversion to digital code is thu~ converted by 59 into its logarithm and is then written into the store 61 as its logarithm in digital code~ The ad-dress in the stor~ is selected by address selector 60.
With the completion of logarithmic storage in unit 51 data is drawn from such storage by the processor unit 52.
The nature and operation of this unit is fully described in the aforementioned cognated Patent Application. The technique described therein for the processing by the unit 52 can be described as that of producing a corrected layergram. The unit withdraws data from storage in 61 in parallel sets as earlier referred to by means of addr~ss selector 62, and processes ~L073~Z~
these sets simultaneously, each set being processed in a term-by-term manner in a data processor 63. As the processing of each set takes place the processed data is stored term-by-term in a processed data store 64 in unit 53, the store having dif-ferent sections each for accepting the data deriving Erom one respective set only.
The unit 5~, for accepting stored data from unit 53, includes a so-called output matric store 65 in which the data, when all processing is complete, is held in a form in which it ~;
directly represents the distribution of absorption coefficient over the area of cross-section examined. The addresses of the store correspond to the meshes of a, for example, Cartesian meshwork, each mesh representing directly a particular elemen-tal area of the cross-section examined, and all the meshes together extending without discontinuity so as to include all, at least, of the area of interest in the e~amined cross-section.
At the address of each mesh there is finally stored a signal which represents, to the degree of accuracy permitted by the equipment, the absorption coefficient of the material of the ; 20 body lying within the elemental area of the mesh concerned.
When the storage is complete for all meshes, the image may be -~
displayed for example by cathode ray tube or by print out, or again either in addition or as an alternative, may be trans-ferred to magnetic tape stora~e. For any selected one of th~se purposes, or any selected combination of them, the unit 55 functions in accordance with common usage to withdraw data from the meshwork store 65, and use it for the display selec-ted.
As a high degree of accuracy is required in the image reconstruction, interpolation is performed in the unit 54 by means of an interpolator 66 transferring the processed data stored in the respective stores 6~ of unit 53 to the output matrix store 65. The interpolation is achieved by co-operation ~73~Z~L
between an address selector 67 and a beam path data store 68 as described in the aforementioned cognated patent application.
In the apparatus being described the angular separa-tion of the narrow beams, which are lmm wide, is 2/15 of a degree, and output signals are derived from the detectors after each angular displacement of the source 5 about the centre O of

2/15 of a degree. Afte~ each increment of rotation of this magnitude, each narrow beam will assume a position which is parallel to the posi~ion which one of its neighbours occupied prior to this increment of rotation.
It is therefore possible by suitable timed selection to assemble beam absorption data signals for sets of parallel beams. Selection of this nature could produce signals corres-ponding to parallel sets of beams angularly separated by 2/5 o~ a degree. However the processiny used in this example is arranged to provide such sets at 2/3 degree separation. Thi~
will be described in more detail hexeinafter~
The signal processing system used in conjunction with the present invention isthe convolution method which is descri-bed in differing forms in the aforementioned co-pending Canadian patent application. The technique essentially consists of arranging the exploring beams in groups related to zones which are concentric with a point for which an absorption val~le is to be calculated. These groups are chosen such that a first group passes through all such zones, a second group passes ~ -through all but the central zone, a third group passes through all but the innermost two zones and so on. The absorptions of the beams in each group are then totalled for the group and multiplied by respective zonal ~actors, known as "L-factors". The sum of the totals, as thus weighted, is proprotional to the ab-sorptian of ~he material in the plane examined ancl at the chosen evaluation point. A plurality of such values--for a suitable number of evaluation points i~ then used to build up the desired image -16-~73~L;~

In the aforementioned Canadian application only the case of exploring beams all of which are of equal width was considered in relation to the underlying principles of proces-sing acquired beam absorption data to yiel:d a useful image reconstruction. The techniques outlinedL may be applied to the narrow-width beams described with reference to Figure 9. ~ -~
With regard to the broader beams that ha~e been refer-red to, such beams are utilized for the purpose of adding in small final corrections, and procedures of great exactitude do not require to be app~ied to them. One method of treating them is to regard each as a contiguous set of fine beams, apportio-ning the broad beam absorption equally among the notional fine beams. Alternatively, as mentioned hereinbefore a single fine beam may be used to obtain an absorption value which can be allocated to each such notional beam. This single fine beam may be conveniently placed in a position which would have been central to the equivalent broad beam. The fine beam series of L-factors is then extended to include the notional fine heams.
If the multiplication of absorption values by L-factors can be performed very rapidly, by for example a special form of cir-cuit such as is described the co-pending cognated application, then this procedure may be adopted. On the other hand, with slower methods of computation, in processing can be saved by assigning special L-factors to broad zone corresponding to ~ broad beams.
;~ A particular case in illustration of this technique is afforded by the situation in which the point in the cross section at which the absorption is required to be evaluated is on the axis of orbital rotation~ In this event, and refer-ring to Figure 7, the annular zones corresponding to fine beam width extend out from the point O as far as the circle 32.

The next surrounding zone is one of width equal to the distance between the bounds 47 and 48. The next further surrounding ~ 3~2î ~
zone is one of width e~ual to the distance between bounds 48 and 49, and finally there is a zone, surrounding all, of width which i5 the distance of bound 50 from bound 49.
Considering first the innermost broad zone this may be thought of as three narrow zones concentric about the point O that continue the sequence of fine beam zones envisaged as extending as far as the circle 32. In this notional sense the fine beam sequence of L-factors is correspondingly extended.
Then however rather than use these latter factors directly their average value is employed to multiply the innermost broad beam absorption, and is thus taken to be the L-factor appropri-ate to the first broad zone.
On the same form of procedure an L-factor is also assigned to the next surrounding broad zone, and in the same way a corresponding L-factor is determined for the final broad zone. As an example a typical value of the L-factor for the first broad æone is 0.001, while that for the next zone is 0.0006, and that for the final zone is 0.00005. For the rea-son that the degree of correction effected by the broad beam is small only the broad beam L-factors do not require to be determined with great precision. Moreover, with few broad beams to take into account it is not difficult to find L-factor values for broad zones by the process, if desired, of trial and error.
The situation is not quite so simple in general as in ;
the particular case just considered, that is to say when the evaluation point is not on the orbital axis. The procedure then, which still allows overall of a useful saving in time of processing may be explained most simply by assuming the equiva-lent L-factor multiplication procedure that is adopted in prac-tice for convenience, and which will be referred to again in more detail. As it has been explained thus far the L-factor multiplication procedure is one in which beam absorptions are ~073~L2~l summed in zones and the absorption sums each multiplied by the corresponding L-factor, all sums so weighted then being added. ~ -It is an equivalent procedure to multiply out the L-factor series not on the zonal basis as just stated, but taking one parallel set of absorption data at a time to multiply out the L-factor series with the absorption values of the set in an otherwise identical manner. It is necessary then to store the multiplication products in intermediate storeO For the present, however, it is sufficient to consider that in proceeding with a parallel set there will be linear intervals corresponding to the zonal intervals and equal to the zonal widths, with respect to which linear intervals the L-factors are now distributed rather than with respect to zones. With the introduction of broad beam L-factors, these factors will be associated with -;-broad beam intervals, just as fine beam L-factors will be asso~
ciaked with fine beam intervals.
It will be evident that in the multiplication with a parallel set it can happen that a fine beam series of intervals overlies entirely a broad beam. In this circumstance the broad beam absorption datum is resolved into a sequence of notional fine beam absorption data of equal value totalling the broad beam datum. If the fine beam series only partially overlies a broad beam then the broad beam is resolved only to this partial extent with respect to the fine beam series for fine beam multiplication leaving over a residual and adjacent notional broad beam. The absorption value to be associated wlth this notional broad beam is appropriately used in broad beam L-factor multiplication. For example the beam absorption value may be added to another constructed value deriving from an adja-cent broad beam, the sum value being multiplied by a correspon-ding broad beam L-factor. Where a broad beam interval falls over fine beams, the fine beam data are summed to construct the absorption appropriate to a notional broad beam of the -19~

3~073~21 width of the interval, and this absorption is then multiplied by the L-factor for the interval. On these lines the absorp-tion can be evaluated with respect -to any point in the examined cross-section, and with a saving in processing time as compared with processing of purely fine beam type.
As will be realized whether broad beam L-factors are made use of in the processing, or not, t;he utilization of broad beams in themselves is valuable for the reason of the saving they permit in the number of scintillation crystals and corres-ponding photo multipliers.
In regard to the scheme of beams considered with res-pect to Figure 7 it is to be noted that/ apart from their paral-lelism with one another, the beams have implicitly been treated as though they were of uniform width. However, in the apparatus of Figures l and 2, the beams defined by the detectors are not of this character, being narrower on one side of the explored field and wider on the other. The effect of this disparity is minimized in the apparatus described by not restricting the orbital motion to the theoretical range of 180 degrees, but by allowing it ~o continue for 360 degrees so that for every beam disposition of the first 180 degrees of scan there is a second which is identical except for the fact that the direction of passage of radiation is reversed, and with it the sense of the disparity. The average of the two beam absorptions is then taken to produce data corresponding to a beam of virtually uni-form width.
A further extension of the orbital motion is also used to reduce the number of scintillation crystals and corres-ponding photo-multipliers used in connection with the large number of narrow-width baams. Thus the number of paixs of these crystals and photo-multipliers is halved by leaving a gap of one beam width between each successive pair, and the gaps con-sequently left in the group of narrow-width beams are made good , . . : ..
::. . :.. :. ..

` ~73~1lZ~

by a lateral displacement of the X-ray source and detector means to the extent of one narrow beam. A further orbital rotation of 360 degrees then provides the missing information.
~his is illustrated in Figure 9 which shows at 5 and 18 the positions of the X-ray source and the collimators for the scin- !
tillators during a first of two orbital rotations. This figure also shows at 51 and 131 the positions of the X-ray source and the collimators during the second of the two orbital rotations, showing by dotte~ lines the result of displacement of the beams to fill the gap between the beams indicated by the full lines.
This technique may however be dispensed~with by using close packing of photo-multipliers and crystal or by using multi ; channel devices. Conversely the tec~mique can be extended to make good the omission of further crystals and photo~multipliers.
A three revolution method may for example be employed. However, if the apparatus is to be used or e~amining the body of a pati-ent in regions where the breathing of the patient can cause undesired body movement in the cross-section of examination, unless the patient temporarily withholds his breathing, it is desirable that the time of orbiting should be brief. The num-ber of possible revolutions of orbiting thus tends to be strict-ly limited. The technique shown in Figure 9 forms part of the ;~
subject matter of co-pending Canadian Patent Application No.
204,299 dated July 8, 1974 in the name of G. N. ~Iouns~ield.
Having decided upon the use of a particular series of L factors and assuming the acquired beam absorption available in logarithmic form as earlier described, and moreover available in the form of parallel sets, the processing to be performed by the unit 131 of Yigure 8 may be accomplished by means of an appropriately programmed computer or the special circuits de-scribed in the aforementioned Canadian patent application No.
198,145.
As previously indicated, and in relation to the fine ....

733LZ~

exploring beams, the angular interval between the equally angularly spaced parallel sets of data is chosen to be 2/3 degree. Considering detectors spaced at intervals of 2/3 de-gree about the orbital axis; it will be evident that the readings of successi~e detectors will relate to a parallel set if they are taken after successive move~ents of 2/3 degree.
Another parallel set may be started, from the first detector, after the first movement of 2/3 degree, this second set being inclined at 2/3 degree to the first set and so on, yielding sets at all the re~uired angles. With the relatively narrow beam widths emplcyed in the apparatus, four detectors are how-ever interposed between each pair disposed at 2/3 degree spa-cing, the interposition being such as to give a 2/15 degree interval between detectors. Consider now Figure lO, which illustrates the group o~ components denoted by unit 51 in Figure 8 in more detail. ~ll detectors included in means 18 are regarded as falling into one or other of five different categories. Category l may be regarded as that of the initial sequence of detectors between which the added four detectors are interposed. Category 1 detectors are thus the first of successive groups of five detectors. Detectors of category 2 are the second of these groups; category 3 the third, and so on. The outputs of the detectors of different categories are then sampled at different times according to their category.
Detectors of category 2 are sa~pled at time T later than those of category l, while those of category 3 are sampled at time 2T
later, and so on, the sampling cycle occupying time 5T, this time ~eing the duration of scanning of the orbital motion over 2/3 degree.
It has been seen that, using the data from those detectors now classed as detectors of category l, parallel sets of data can be constructed with an angular inter~-al of 2/3 degree. In any set so constructed the data corresponds to ., .;,.,. :. . .. ,.;

73~

beams spaced apart by intervals of extent such as to accomo-date the spread of four intervening beams at the spacing of fine beams. Data corresponding to the locations of such inter-vening beams is derived by means of the sampling of the outputs of the detectors of categories 2, 3, 4 and 5 so as to provide full sets.
In general, beams of a fan having an angular spacing -~
of ~ may be combined in n categories to provide parallel sets at na angular separation. However each such set will have _ times the number of beams provided, at the same separation, by a fan of beams of spacing na. In the case described hereinbe-fore n is given as 1 and a is 2/15 degree. Therefore na is also 2/15 degree. In the example relating to Figure 10, n = 5 and na is therefore 2/3 degree.
The duration of sampling, at any time of sampling detector outputs, is such as to correspond to the ine beam spacing, allowing for laperture effect' to spread the effective radiation density distribution across the beam so that the over-all effective spread o the beam is twice the spacing of beams of a finally derived parallel set. Sampling is effected by causing the Miller integrators 52 earlier mentioned to start and cease integrating at appropriate times, and deriving their inte-grated outputs. The Miller integrators are thus used in their known role o analogue stores in which they sample and hold, thereafter to be reset so as to be available for further samp-ling in the same manner. The derived parallel sets of data as they are derived are stored in respective stores so as to be dixectly available for convolution.
In Figure 10 output conductors 70 proceding from the detector means 18 carry khe outputs of those detectors giving narrow-beam absorption information, one conductor being associ-ated with each such detector. The conductors 70 are shown sepa-rated out according to the respective detector categories 1, 2, ~3 ;

.. ~ . , .

1~73~
3, 4, 5. As so classified the conductors proceed into the initial processin~ and storage unit 51.
The outputs of the broad beam detectors correspond respectively to the notional fine beams into which the broad exploring beams may be regarded as resol~ed. For e~ample a broad beam of 10mm width is re~arded as resolved into 10 no-tional fine beams. In principle the detlector sensing the absorption suffered by the broad beam thus feeds OUtpl1tS in-to 10 separate fine beam channels, a separate output conductor applying signals to each channel. Since however each such sig-nal must be the same, on the grounds that there is no reason to apportion the detector output other than equally, in practice each broad beam detector feeds one output conduc~or only and one corresponding channel. On this understanding the operation of these channels will become clear with the further descrip-tion of the e~uipment in relation to the outputs of the detec-tors giving information concerning the absorption of the fine exploring beams. For simplicity therefore detector output conductors relative to the broad exploring beams are not shown in Figure 10, and for the same reason only one typical fine exploring beam channel is illustrated in the figure. This corresponds to the typical fine beam detector output conductor 70.
After initial processing/ immediately to be described, the data is distributed to the sections 1, 2, 3. ...., _ of store 61 ~Figure 10) to which the data deriving from all other conductors is correspondingly communicated, so that in each store there is held the data of one parallel set, o~e store being used for each set.
Considering the t~pical conductor, designated 70k in the figure, the currents carried by the conductor are applied to the input of the gain controlled amplified 56k. The gain of this amplifier, as indicated earlier, is adjustable so that D'73~1L2~
the relative sensitivities of the various detectors may be compensated and so that variations in the emission from the X-ray source may also be compensated. The gain control may also provide means, if desired, for compe~sating for drift in the relative sensitivities in the course of scanning. The various amplifiers such as 56k are controlled in gain from gain control unit 71.
The output of the amplifier 56k is fed to the analo-gue store 57k which as stated earlier takes the known form of Miller integrator in sample-and-hold form. Sampling by all such circuits as 57k is under the timing control of timing unit - 72, which also controls the time of read out and of reset of these circuits. The read out from circuit 57k is converted from analogue to digital form by the circuit 58k and fed to distributor circuit 60k which communicates with the various sections 1, 2, 3, ... n of store 61. In the event that all the sampled data deriving from the various detectors related to parallel exploring beams, which :is not in fact the case, since the exploring beams have the divergence of the radiation fan, all the distributors of which 60k is typical would distribute at any one time to one store section only, corresponding to one particular angle of the orbiting scan. This store would then be completely filled at this time, disregarding for sim-plicity the developments described to take account of the non-parallelism of individual exploxing beams, and to reduce the number of detectors employed by a factor of two. These mea-sures as earlier noted lead to a two revolution scan rather than one of 180 degrees, the latter being all that is necessary in principle. However, no store section is filled in one single filling procedure even if the measures adopted reduce the appa ; ratus to one of a simple 180 degree scan. Rather, contributions to a given parallel set are made over a range o~ sampling times in the way enunciated from a range of different cletectors.

- . ~ . . ......................... , - . , .
.. ..

` 1~731Zl Following such a programme of timing th~ distrubutor circuits such as 60k contribute to the parallel set stora~e under the - control of unit 72.
The parallel set data so set up in the store sections 1, 2, 3, ...., n is available to be passed to the convolution processing unit after logarithmic conversion. To accomplish this conversion and as data is established in the addresses of the parallel set stores it is fed to the logarithmic converter unit 59 to be written back into the same address from which it was withdrawn, but in logarithmic form. Thi~ is performed un-der the timi~g control of the unit 72. It is to be noted that in the form of apparatus described every address receives two contributions, one corresponding to one direction of transmis-sion of the relevant beam, and one to the beam in 180 degree relation of scan. The data at an address is not complete until both such contributions have been made and logarithmic conver-sion cannot be effected until that time.
It will be understood that the present invention may be applied to any scanning arrangement suitable for apparatus : 20 of the type described in the aforementioned specification and applications, in particular a linear scan superimposed as an orbital scan. As a further example, considering the technique described for selecting sets of parallel béams from a larger set at a variety of angular dispositions, other methods are known for assembling such a larget set. In one method a fan shaped distribution of beams is given a linear scan and further orbited to repeat that scan at a variety of angles. However it has been found khat extreme positions of the linear scan may not provide enough individual beams to complete all of the parallel sets. In such cases, according to the invention, ab-sorption information required for a disposition for which a beam is missing may be supplied by any other beam lying sufficiently near to the required disposition.

~ . .

.:.: .. ~ . .. . ... .... "... ....

: "
10731Zl ~

Furthermore thP invention may be combined with any suitable signal processing system.

:' ' ~ .
`,f ~`' `

; ` '''''' : ?
.

'~',:

~' .

~ -27-

Claims

THE EMBODIMENTS OF THE INVENTION IN WHICH AN EXCLUSIVE
PROPERTY OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS.

Apparatus, for examining a body by means of pene-trating radiation, such as X-radiation, comprising a source of a fan-shaped distribution of said radiation arranged to pro-ject said radiation along a plurality of paths through a slice of the body, means for scanning said source relative to said body to project said radiation through said slice along further paths, detector means for providing output signals indicative of the amount of absorption suffered by said radiation on tra-versing said paths, said detector means including a plurality of detectors, adjacent ones of which are disposed to receive radiation along respective paths inclined to each other at a given angle, and means being provided for sampling said detec-tors in groups at interleaved times to derive therefrom output signals relating to sets of substantially parallel paths, neighbouring sets being inclined to each other at an angle greater than said given angle.