JPS6411296B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method and apparatus for inspecting objects with radiation, such as X-rays or gamma rays.

The method and apparatus according to the invention can be applied to images such as images on a cathode ray tube or other image forming device, photographs of such images, or maps of absorption coefficients that can be generated and subsequently contoured by a digital computer. Any form of radiation can be used to facilitate the acquisition of radiation.

Patent Application No. 1973-66087 (Special Publication No. 1274, Showa 52)
In the method and apparatus for inspecting an object described in , radiation from an external radiation source is passed in the form of a pencil beam through a portion of the object to be inspected. The beam is superimposed with a scanning motion so that it assumes a number of different positions, and a detector is used to provide a measurement of the absorption of the beam at each position after the beam has passed through the object. ing. As the beam assumes these various positions, the radiation source and detector are caused to reciprocate in a plane and to orbit about an axis perpendicular to that plane.
Their various positions lie in a plane in the object under examination in which, by processing the beam absorption data provided by the detector, a distribution of absorption coefficients for the radiation being used is obtained. The processing is such that the absorption distribution finally displayed is the result of a series of approximate calculations.

The method and apparatus described in the above-mentioned patent application has been found to be quite satisfactory for obtaining cross-sectional representations of parts of the human body, such as the head.

Special Patent Application No. 1983-63812 (Special Publication No. 1988-8821)
describes an apparatus for obtaining beam absorption data relatively quickly. According to this patent application, the data is derived by transmitting a fan of X-rays through the object in the plane to be examined and placing an array of detectors on the opposite side of the object. by measuring the x-rays transmitted along a set of beam paths within the x-ray fan. The set of passageways extends over an angle sufficient to encompass the entire area of interest in the plane of examination, so that the radiation source and detector can be simply orbited around the object under examination. A complete scan can be done.

According to the invention, a radiation source means adapted to irradiate an object to be inspected with penetrating radiation; detector means comprising a plurality of detectors adapted to detect and compensating means for reducing the effect of differences in sensitivity of different detectors, said compensating means comprising said radiation source means and said radiation source means; displacement means for producing a relative displacement between the radiation source means and the detector means; and a displacement means adapted to receive an output signal from the detector and wherein the radiation source means and the detector means are connected to the first and second radiation sources. means adapted to predict the value of the output signal to be obtained from an adjacent detector using the output signals obtained from the series of detectors while taking the relative positions of the detectors; and means adapted to utilize the determined values to at least partially compensate for differences in detector sensitivity.

The means for predicting the value of the output signal to be obtained from the adjacent detector includes means for deriving a signal indicative of the difference in absorption of the object to be examined relative to the radiation transmitted through the object. can be included.

Another object of the invention is to utilize the difference signal in reconstructing a representation of the varying absorption of said plane of the object to be examined, and according to another aspect of the invention, radiation source means adapted to irradiate the body; detector means adapted to detect the radiation after the radiation has passed through the body; means for causing the radiation source and detector means to scan with respect to the object to be examined to permit detection of radiation passing through the object along the object (at least some of which are arranged to intersect with each other within the object); and obtained from said detector means,
means for processing (according to a convolution method) an output signal indicative of the absorbed absorption of said radiation along each of said paths, said processing means processing a signal representative of the difference between said signals; A radiographic apparatus is provided that is adapted to operate.

Embodiments of the present invention will be described in detail below with reference to the drawings.

Referring to FIGS. 1a and 1b, a test object 1 is supported on a suitable base plate 2 facing upward. The base plate 2 is supported at both ends by support members 3 and 4 fixed to the base 5. The support member 4 indirectly supports the end of the base plate 2 via the drum 6 and the shaft 7, and the shaft 7 is attached to an L-shaped member 8 that supports the base plate 2.

A detector 10 in line with the X-ray tube 9 is mounted on the outside of a cylindrical drum 13 by respective supports 11 and 12, which drum 13 supports a portion of the object 1 to be inspected and is connected to a motor 14. Therefore, it is designed to be rotated around the object to be inspected. The motor 14 has teeth that engage teeth on a gear ring 16 formed on the outside of the drum 13. drum 1
The closed end of 3 centrally carries a rotatable bearing 17 on a shaft 7, which shaft 7 is aligned with the longitudinal axis of the drum 13 and which extends the axis of rotation of the system as a whole. It consists of drum 1
The open ends of 3 are supported on bearings 18 (only one of which is shown in FIG. 1a).
The bearing 18 is supported by the support member 3.
Power to the x-ray tube 9 and electrical signals from the detector are transmitted by cables 19 and 20, respectively, which extend along the outside of the drum 13 and are connected to suitable cable guiding members 21, 22 and extend to the drum 6, respectively. In this case, the drum 6 is the drum 1
It carries enough turns of the cable to make one complete revolution. With this, drum 13
It will be appreciated that must be returned to the initial position after each scan.

The patient is positioned such that the part of his body to be examined is in the path of the X-rays from the X-ray tube 9 to the detector 10. The X-ray tube 9 includes collimator means for causing the radiation to be emitted in the form of a planar sector trajectory 23 (see in particular FIG. 1b), the dimensions of which sector are as shown in FIG. 1a. perpendicular to the plane of the page of FIG. 1b and parallel to the plane of the page of FIG. 1b. The area of interest of the object to be inspected is the object 1 to be inspected.
The material 25 is surrounded by a positioning ring 24 containing a material 25 having absorption properties similar to that of the object 1, which is placed between the ring 24 and the object 1 to be inspected in order to expel as much air as possible from the vicinity of the object 1. It is tightly crimped.

Therefore, when the motor 14 is energized, the drum 1
It will be appreciated that 3 is rotated around the object under examination while still carrying the X-ray tube 9 and detector 10. In this case, the fan beam 23 is rotated around the object to be examined in a fixed plane, so that the object to be examined is irradiated from several different directions in that plane.

As clearly shown in FIG. 1b, N detectors are disposed opposite the X-ray tube 9 and extend across the width dimension of the fan beam 23. distributed. Each detector includes a collimator 26 (not individually shown) such that if the fan beam were stationary, it would only receive radiation that passed through each straight path within the examined body. The width of the passage is regulated by the dimensions of the collimator. In practice, the fan beam is rotated steadily around the object under test, and the output signal from the detector needs to be integrated over a short finite time to obtain a reliable reading. , the signal obtained from a given detector relates to a path that is expanded to some extent with respect to the path fixed by each collimator. The fan beam 23 can be rotated by an angle between 1° and 2° with respect to the object to be inspected during each finite integration time, which angle is 1° in the present example.

A typical detector consists of a scintillator crystal that is arranged to receive x-rays and generates a visible output signal in response to the radiation, the visible output signal indicating the amount of radiation incident on the crystal. It is transmitted to a secondary electron multiplier tube which generates an electrical output signal for display. These output signals are processed in a suitable manner to produce an indication of the change in absorption for radiation in a direction transverse to the plane of the object being examined.

As mentioned above, when output signals obtained from several different detectors are processed, inherent differences in detector performance will result in undesirable patterns being superimposed on the display.

The shape and strength of those patterns depends in part on the type of processing used to generate the display. To practicably eliminate such undesirable patterns from the display, the present invention provides relative motion between the fan beam 23 and the detector 10. This means that when the output signal from a certain detector (for example, the k'th detector) is known at a certain point in time, the performance of the next adjacent detector (for example, the k+1'th detector) is It gives information that allows calculation of what the corresponding output signal of this neighboring detector should be, given the same performance of the detector. This allows the calculated value to be compared with the actual output signal recorded at said instant by the k+1'th detector. In this case, the difference between the two compared values can be attributed to the difference in performance between the k'th and k+1'th detectors. By this measure, such differences are taken into account and the intensity of the aforementioned undesirable patterns is substantially reduced. Of course, the kth is compared with the k+1′th, the k+1′th is compared with the k+2′th, and so on.
It will be understood that corrections are made for all detectors by comparing .

In this embodiment, the change in relative position between fan beam 23 and detector 10 is made by displacing the beam source from position Bu to position Bd in FIG. 1b. Although in practice it has been stated that the detector output signal is integrated over a period corresponding to a 1° angular rotation of the fan beam with respect to the object under test, when the present invention is used, the fan beam is not displaced. It is necessary to separate the output signals obtained as they emerge from the position (subscript u) and the displaced position (subscript d) on a time basis. This can be done, for example, by maintaining the fan beam in position Bu for approximately half of the aforementioned period and then displacing the beam source to position Bd for the other half of that period. Ru. The output signals obtained from the detector during each half period may be separately integrated and provided to each memory. Since the steady rotation of the drum 13 proceeds unhindered,
At the end of the second half-period, it is necessary to return the fan beam source to the position corresponding to Bu, but of course by that time the fan beam is at the previously mentioned 1° angle from the position shown in Figure 1b. You can see that it has been displaced by The process of displacing the fan beam source from one position to another and back again is repeated for each degree of rotation.

Alternatively, it is possible to oscillate the position of the fan beam source more rapidly so that each position is occupied several times during an increment of angular movement. However, in the first embodiment of the invention described below, it is assumed that the last method described above is adopted.

As is clear from the above description, the rotation of the drum 13 around the object to be inspected needs to be accurately monitored. In this embodiment, such monitoring is accomplished by providing a scale in the form of opaque radial lines on a disc 70 of transparent material driven by motor 14. The disc rotates with respect to a transducer unit 71 containing a light source and a photocell;
In this case, said photocell is connected to a suitable circuit of the known type for generating timing pulses, a pulse being generated each time the path of light to said photocell is interrupted by one of said graduation lines. Ru. The distance between the radial lines on the disk 70 is such that adjacent ones of the pulses 72 are 1/2 the width of the drum 13.
They are chosen to be separated by a time equivalent to a rotation of °.

Referring now to FIG. 2, the numeral 13 designates the drum member of a scanning or beam absorption data acquisition unit as described in connection with FIGS. 1a and 1b. The fan-shaped beam 23 extends over the fan-shaped portion in the plane of the paper of FIG. 2 between the limit lines indicated by 27 and 28, and is believed to originate from an effective point source at Bu. This fan-shaped radiation passes through the object to be inspected and later irradiates onto the array of detectors 10. According to one aspect of the invention, as described below, the relative positions of the X-ray source and the array of detectors 10 are displaced so that a fan-shaped radiation probe due to differences in sensitivity between the detectors is provided. It is provided that defects in the final image reconstruction of the absorption distribution in the cross-sectional part of the object to be examined can also be reduced in a manner that will be explained later. As will be seen below, the relative positional displacement is accomplished by deflecting the excitation electron beam of the x-ray tube used to generate the search radiation. Deflecting the electron beam in this manner causes the x-rays to be emitted from a patch (on the x-ray tube target) that is at a different location than the patch that would emit x-rays if the electron beam were not deflected. In FIG. 2, this patch (which is so small that it can be considered for operational purposes as a point radiation source) is schematically shown as being displaced from position Bu to an adjacent position Bd, where lines 27 and 28 are The displacements resulting in lines 27' and 28' are not large, but are shown exaggerated in the drawing.

Although the detectors are not shown individually in Figure 2, as mentioned above, they each consist of a scintillation crystal that can be impinged by radiation, and that crystal is a secondary electron multiplier. is optically coupled to. In this case, the incident
Excitation of the crystal by the line generates light that is coupled to each secondary electron multiplier to produce an output current from that secondary electron multiplier. The current becomes the output current of the detector. The output currents of the various detectors flow along the conductors (indicated by 29 in FIG. 2) constituting the various cores of the cable 20 (FIG. 1a), in this case the kth The detector supplies its output current through conductor 29k to a signal amplifier 30k associated with the detector. amplifier 30
has individual gain adjustment means of known type to compensate for gross inequalities in detector sensitivity. The gain of the amplifier 30 is controlled by a gain control unit 31.

The output of amplifier 30k is fed to analog storage 32k, which in this embodiment is a Miller integrator of the known type. Integrator 32k receives timing pulses 72 from timing units 70, 71, shown as blocks in FIG. 2, each of which reads and resets the integrator in a known manner. The integrator circuit 32k is configured so that the fan-shaped beam is connected to the object to be inspected.
The amplified output of the k'th detector is subsequently integrated during the aforementioned time interval, which corresponds to half the time required to rotate through an angle of 1°. this
The angle of 1° is chosen because it corresponds to the angle that one of the detectors 10 makes at the source 9, ie the effective width of the search beam sensed by the detector K. In the analog-to-digital converter 33k, each signal generated upon resetting the integrator is converted to digital form and provided to a logarithmic conversion circuit 34 which converts the signal, although still in digital form, to a logarithmic signal. This logarithmic conversion circuit is common to all converters 33. It is operated under the control of timing pulses from units 70, 71 and is adapted to store the converted signal for the required time 33.
Take the digital signals provided by all converters such as k. All circuit functions represented by blocks 34 onward in FIG. 2 are provided by a suitably programmed digital computer, the structural details of which are not directly relevant to the present invention. Since there is none, I will omit it. The logarithmic converter 34 supplies the signal converted into logarithmic form to the distribution circuit 35 . This distribution circuit sends the signal to the digital storage 36 if the signal relates to the position Bu of the x-ray source, so that the corresponding position of the x-ray source is
If it is Bd, it is sent to the second digital storage 37. The switching pulses for controlling the distribution circuit 35 are synchronized with the pulses of the timing circuits 70 and 71 described above, which timing circuits generate a substantially sinusoidal deflection waveform 73, described below, associated with the radiation tube 9. Position generated by scanning circuit 73
It also generates pulses to control the displacement of the X-ray source between Bu and Bd.

As a result of the acquisition of absorption data by the scanning unit,
Upon completion of the trajectory scan, the storage 36 stores the absorption values obtained as described above and then matched to the matrix pattern.

R ₁₁ R ₂₁ R ₃₁ ……R _k1 R _K+1,1 ……R _N1 R ₁₂ R ₂₂ R ₃₂ ……R _k2 R _k+1,2 ……R _N2 〓 〓 R _1t R _2t R _3t …… RktR _k+1,t ……R _Nt 〓 〓 R _1T R _2T R _3T ……R _kT R _k+1,T ……R _NT In this array, the first row is the detector array for 1/2° rotation the second row corresponds to the third sampling for the third 1/2° rotation,
Lines starting with R ₁ t correspond to samples (2t−1),
And the last row corresponds to sample (2T-1), which is the last sampling of the scan. The first column gives the absorption values obtained from the first detector in the detector array in the time order of sampling that column downward, and the second column gives the absorption values obtained in the same way from the second detector in the detector array. and thus up to the value obtained from the last Nth detector. Generally speaking, R _kt represents the absorption value obtained from the kth detector at the (2t-1)th sampling time.

Correspondingly, in the memory 37, different absorption _values are stored in the same pattern, typically for the represents the value associated with the detector. Due to the alternation of the radiation source positions, the value R' _kt will have an associated sampling time intermediate that of R _kt and R _k,t+1 .

Block 38 retrieves absorption data values from storage in stores 36 and 37 to obtain corrected values that are less affected by differential detector sensitivity errors than the data held in those stores 36 and 37. Processing circuitry that uses those absorbed data values is represented. The data transferred to the storage device 39 is processed by convolution using any suitable method, such as that described in Japanese Patent Application No. 49-47032 (Japanese Patent Publication No. 59-11150). . First, however, the data is sorted by distribution circuit 40 into data relating to parallel sets of radiation paths through the object. Each of these parallel sets is transferred to each section of the parallel set storage 41, but
There are n such sections, n having a value smaller than the number N of detectors in the detector array.

Next, the nature of data correction will be explained with reference to FIG. In this figure, circle C represents the orbital path between the radiation source and the detector in the detector array. For the sake of simplicity, the orbital paths of the various detectors can be considered to be similar, although not necessarily identical, to the orbital path of the radiation source, and the arrangement of the detectors is relatively similar to that of the radiation source. It is assumed that they form a small angle. The center of the circle C representing the axis of the orbital rotation is indicated by 0 in the figure, and the undisplaced position Bu of the radiation source at the averaging time of the tth sampling is also indicated in the figure. If the radiation source is then displaced, the position of the radiation source at the average time of the next sampling is
Indicated by Bd. This figure shows the radiation from the radiation source at the time of sampling both when passing through the search area and when corresponding to the k'th detector at each time point. Thus, path P is the path of the radiation in the undisplaced position of the radiation source, and P' is that in the displaced position. In the first position, the radiation path is from the orbital axis.
OX apart, and in the second position,
They are separated by XO′.

Consider any radiation path such as P or P′,
Let it be separated by r from the orbital axis and inclined at an angle O to the reference line in the exploration plane. The total absorption of radiation along the path between the radiation source and a detector, such as the kth detector, which intercepts the radiation and determines the absorption, is a function of the magnitude of both r and θ. Will. Thus, if f is the actual value of the absorption (which is clear from some kind of determination that must contain errors), then f can be expressed according to the formula f=f(r,θ) . If r increases by Δr and θ increases by Δθ, the value of f will increase by Δf=∂f/∂rΔr+∂f/∂θΔθ. This formula can be written as:

∂f/∂r=df/dr−∂f/∂θ dθ/dr The meaning of this result is that ∂f/∂r is adjacent to a certain path and the absorption along that path is known. and is a quantity that can represent the absorption along a path parallel to it. The result also has the implication that ∂f/∂r can be determined from data of the type seen by the apparatus described in connection with FIGS. 1a, 1b and 2. ing. It thus makes it possible to ascertain the relative error in detector sensitivity between these two adjacent paths, as required for more accurate operation of the device, and thus eliminates poorly compensated differences in the detector. Reduces artifacts that tend to appear in the final image reconstruction due to sensitivity.

Applying these principles to the tth sampling by the kth detector, the value
R _kt is stored in the memory 36, and R′ _kt is stored in the memory 37.
When stored in , the following equation holds true.

df/dr | _kt = R′ _kt −R _kt /s where s is the half displacement of the radiation source between the positions Bu and Bd in Fig. 3, i.e. the difference in length between the two perpendicular lines OX′ and OX. be. moreover,
In connection with the two samplings, the following equation holds.

dθ/dr=2s/D/s=2/D Let D be the length of paths P and P', which lengths are considered to be identical to the appreciable limit. Furthermore, if α is the orbital rotation that occurs between samplings with respect to values R _kt and R _k,t+1 , then
The following formula is obtained.

∂f/∂θ | _kt = R _k,t+1 −R _kt /α These results give the following equation.

∂f/∂r｜ _kt = df/dr｜ _kt −∂f/∂θ｜ _kt dθ/
dr The terms on the right hand side of this equation are determined by the verified data on the operation of the device and according to the relationships derived with respect to them.

Thus, it can be written as follows.

R ^* _k+1,t = R _kt + ∂f/∂r _kt・s+∂f/∂θ _kt・(α−β) Note that R ^* _k+1,t is the (k+1)′th detector is the absorption value that should be given at the t-th sampling assuming that the detector has the same effective sensitivity as the k-th detector. The last term contains a small correction related to the angle β between the paths from the radiation source to the k'th and (k+1)'th detectors, but if β is small enough, as is the usual case. For example, this correction can be ignored.

Applying these comparison methods between adjacent detectors, the data for any row of absorption values, i.e. for any set of such values obtained at any sampling time, will be determined if all measurements are taken at the beginning of the detector array. It can be corrected to give a measurement as if it were made with an effective detector sensitivity equal to the detector sensitivity. Of course, it is necessary that the absorption value for each detector used in the correction for the next subsequent detector is itself a corrected value. Therefore,
These corrections must be made sequentially.

In accordance with the above-mentioned idea, the processing unit 38 retrieves the stored absorption values held in the memories 36 and 37 and thus calculates partial differences of the type ∂f/∂r and ∂f/∂θ. Processing these values on the equations derived for the coefficients, it is determined that for every row set of R values held in memory 36, a detector has the same effective sensitivity as its neighbor. calculate a set of R ^* values representing (within measurement error) the absorption determination that would have been given by that detector if This corrected determined value is given to the memory 39. However, the calculator may make the correction in different ways. It will be appreciated that R* _k+1,t /R _k+1,t is a measure of the sensitivity of detector k+1 at time t to detector k at the same time. All these detector sensitivities can be stored. Here, it can be assumed that the difference in effective sensitivity of the detectors is the same throughout the data acquisition period and that an average representation of relative sensitivity can be determined using all row sets of R data. Once the average value of relative effective sensitivity is so made available, the R value in store 36 can be transferred to store 39 and corrected by processor 38 in accordance with the average relative effective sensitivity determination.

The set of R' values held in memory 37 may also be corrected by any of the methods described above, with the corrected values having the advantage of counteracting errors resulting from static errors in data acquisition. in the same manner as the image reconstruction means, in which the set of corrected R values is processed for a second image reconstruction, which can be superimposed on the first image reconstruction to establish an image reconstruction with Therefore, it can be used.

In situations where fluctuations in the effective detector sensitivity over the course of the data acquisition period cannot be ignored, the method of averaging the relative sensitivity cannot be used;
and the R value can be corrected for rows, but for all rows (unless there are steps to the contrary)
will tend to have different uncompensated errors than those of other rows as a result of said drift. This tendency can be alleviated by configuring the extreme detectors in the detector array to measure absorption on a reference path rather than absorption in a path through the test object. According to such an arrangement, the correction to the reference path reference value has the advantage that the reference value will not be averaged over a short sequence of samplings if the drift in this fraction of the data acquisition period is small enough. It can be done. Alternatively, the extreme detectors may be replaced by a group of such detectors whose averaged output is used as a reference.

In a variant of the apparatus described with respect to FIG. 2, instead of having the alternating x-ray source position fixed during the integration period during which the absorption data values are increased, the x-ray source position is alternated at a relatively high speed. At the same time, the amplified secondary electron multiplier output may be appropriately switched between the pair of analog integrating circuits for each detector.
Thus, each R value is generated simultaneously with its R' value, rather than in the sequence described with respect to Figure 1, with the R' value associated with that R value corresponding to the displacement of the radiation source. . In this modified arrangement, the R and R' values are supplied directly to each memory 36 and 37 rather than by sequential switching. However, the basis of the correction is the same principle. In this modified device, the displacement of the X-ray source is preferably rectangular in time, but it may also be sinusoidal if desired.

Ideally, it is preferable to have a rectangular displacement of the X-ray source rather than a sinusoidal displacement, but in practice it is convenient to perform the displacement in very strict accordance with the true rectangular displacement law. You will see that it is possible that there is no such thing. Such precisely performed displacements may entail the use of harmonic deflection components that are disadvantageously high multiples of the fundamental frequency of the alternating motion of the radiation source. If such harmonics did not exist, the transitions between alternate displacement positions would occupy a non-negligible fraction of the alternating cycle, and corresponding to this fraction, the output of the detector would be at either displacement position. It will contain components that do not indicate even a single point but indicate a large number of intermediate positions. In that case, the required corrections will not be made with the same precision;
If a sufficient range of harmonics is used, correction can be made to a satisfactory degree.

It will be appreciated that deflection of the x-ray source can also be achieved by varying the x-ray energy emission from the x-ray source. This could be compensated automatically if the x-ray emissions from the radiation source were monitored by an auxiliary detector and the output of the absorption sampling detector was normalized to that of the auxiliary detector. Dew.

Each of the sampling detectors is provided with collimator means to reduce errors from the radiation scattered by the object under examination, but in principle the collimation can be carried out between alternating positions of the X-ray source. It will be appreciated that it must be such that radiation is incident on the scintillation crystal without discrimination. In fact, if a non-negligible degree of discrimination exists, its magnitude can be taken as the ratio of the sum of rows of R values to the corresponding sum of rows of R′ values; Size is taken into consideration.

Regarding the storage of values in the memory 39, these values need not be the final corrected values for detector sensitivity errors; It may simply be the correction itself, as long as it is the final corrected value of the set minus the initial reference value on which the final corrected value of is based. If those corrections include row-by-row corrections, the entire matrix of such values is such that, during video processing, the average value component is not reached but only gives the desired video reconstruction. It is. In this way, video processing is simplified and
Image reconstruction can be completed by adding the appropriate average magnitude.

The above-mentioned correction principle applied to the transition of the X-ray source occurring in the direction from position Bu to position Bd in FIG. 3 can also be appropriately applied based on the transition from position Bd to position Bu. will be understood. By combining both types of transition-based corrections, the corresponding combination of data from both R and R' values produces an image that is virtually unaffected by detector sensitivity errors. It can be assembled in a single matrix format in storage 39 in preparation for processing.

FIG. 4 shows an x-ray tube configuration in which the relative positions of the x-ray source and radiation detector may be alternated such that the excitation electron beam may be deflected to provide alternating relative positions. being done.

In FIG. 4, numeral 42 indicates the glass tube of the X-ray tube. 43 is a cathode, and 44 is an anode member. In operation, an excited electron beam from the cathode (held at a large negative potential) traverses the evacuated space of the tube, indicated at 45 by a dotted line defining the major dimensions of its cross section. and impinges on a tungsten target 46 provided on anode member 44. This anode member is likewise held at a large positive potential. The cathode 43 has a diameter of 2
It is constructed as a metal spiral with dimensions such as mm and length 12 mm. It is supported at both ends by support wires 47, which pass through pinches 48, through which a heating current can be applied to the spiral for the purpose of electron emission.

When the excited stream of high velocity electrons impinges on target 46, x-ray photons are emitted from the target in all directions. Some of those X-ray photons pass through holes 49 in the anode member, and these sheets of X-ray photons, defined as will be explained in more detail below, constitute a fan sheet of radiation for inspection. . The aperture 49 is closed with a suitable conductive window, as indicated at 50, so that its presence does not alter the electrostatic field to be associated with the anode member 44.

The anode member 51 is supported on the glass tube body 42 of the X-ray tube by a seal shown and the tube 5
The cooling oil supplied through the tube 2 and out the end 53 of the tube provides stronger cooling, particularly in the vicinity of the target 46. In this case, the end portion from which the cooling oil emerges is shaped to widen the flow of the oil.

Electrons emitted from the cathode 43 reach the target 46
As it flows toward the target, it is subjected to a focusing action, and this action is provided by a focusing electrode (not shown in Fig. 4, but shown separately in Fig. 5 for convenience). It is something. FIG. 5 shows cathode 43 viewed from one end, ie, from above in FIG. 4, with support wire 47 shown connected to the cathode spiral. Numeral 54 indicates a focusing electrode bent in one dimension so as to be concave toward target 46. The cathode support wire 47 passes through the cathode 57 and holds the cathode close to the electrode 54 on the target 46 side. Electrode 54 is arranged symmetrically with respect to the cathode. The electrode 54 is also maintained at cathodic potential by connecting it to one of the cathode support wires 47.

A broken line 55 indicates the bending of the equipotential surface formed near the cathode 43 by the focusing electrode 54. This creates a track of electrons emitted from the cathode that is initially spread out but then converges. This focusing effect causes the electron beam to form a strip when it hits the target. The cross-sectional spread of the electron beam is typically 1 mm within the plane of the paper in Figure 5, but approximately 12 mm perpendicular to the plane of the paper, corresponding to the length of the cathode spiral.
It has been extended over the period.

For the purpose of deflecting the electron beam so as to alternate the position of the effective X-ray source, H is shown in FIG.
The magnetic fields shown at 57 and 58 in that figure
It is set by the deflection current flowing through the conductors of the pair of deflection coils shown in . These coils were mentioned briefly above in connection with Figure 2, but they may be of the type used in cathode ray tubes in television receivers, and may simply be of the so-called "hank" type. It can be something. The coils are mounted on a formation 59 that is a sliding fit on the tube 42, and the coils are excited by a scanning waveform such as 74 from a scanning circuit 73 (FIG. 2). The direction of the lines of magnetic force of the magnetic field H generated when the magnetic field H is placed is essentially within the plane of the paper of FIG. With this configuration, the flow of deflection current in the coils 57 and 58 deflects the flow of electrons toward the target 46 in a direction perpendicular to the plane of the drawing. Thus, the impact region of the electrons on the target 46 is deflected in this direction by, for example, 2 mm by the application of the deflection current.

FIG. 6 shows the tube 42 immersed in cooling oil 60 within a lead lined casing 61. The lead backing may have a thickness of 2 mm so as to absorb substantially all radiation emitted thereto. Holes 62 in the lining allow sheet radiation from target 46 to pass through the casing. The cross section of this sheet-like radiation is broken line 6
3, and extends in a fan-shape radiating from its source region in the previously mentioned target 46 in a plane perpendicular to the plane of the paper of FIG. Cooling oil 60 is supplied to hole 6 by suitable window 65.
2 is prevented from passing through. Target 4
Radiation scattered from areas within the casing rather than from 6 is substantially prevented from passing through aperture 62 by lid 64 . This lid, because of the presence of the slot 66 in it, does not affect the emission of the fan-shaped sheet radiation indicated at 63.

Instead of storing the net corrected value in the memory 39 (FIG. 2), it is possible to store the difference between each value and a reference value (such as R ₁₁ ) and perform processing on these difference values. I have already mentioned the good things. It is advantageous to use difference values, since they are already available due to correction processes that have already been carried out. If the process involves a convolution technique,
Although it is still desirable to sort the data into parallel sets using the distributor 40 in FIG. 3, the convolution process itself requires some modification to allow it to deal with difference values as opposed to net values. I need. Assuming that the apparatus of FIG. 2 is adapted to store difference values in a memory 41, these signals are processed by a convolution processing unit 67 as shown in FIG. may be of any suitable form). Furthermore, the data processed by unit 67 is subjected to an interpolation procedure following such processing. This interpolation step is carried out in an interpolation unit 68 to which the data is transferred by 67. Upon interpolation, the data is transferred to a display matrix store 69 where it is maintained in a form suitable for display on a computer printer or cathode ray tube, and that the printer or cathode ray tube display reflects the absorption on the cross section being examined. Show a pattern.

In order to understand more clearly the special nature of the convolution process carried out by unit 67, let the function f(h) represent the line integral of the absorption in the plane of examination, with uniformly spaced values of parameter h. Consider that the samples of that function in values represent the sequence of signals in the set supplied to the memory in FIG. It can be shown firstly and at least effectively that it is possible to reconstruct the pattern of absorption in the examination plane by performing a convolution integral as follows.

In this integration, function q is a defined convolution function, and function q can be chosen such that it gives a predetermined frequency emphasis to the convolved data. The limits ±1/2A are considered to include the entire absorption field in this integral. For convenience, this integral may be divided into a number of partial integrals, each over a range equal to the sampling interval, where the sampled values are multiplied by an appropriate convolution factor corresponding to the partial integral in question. It takes the form of the sum of a finite number of terms.

Figure 8 diagrammatically shows the general properties of the function q(t), whose analytical structure is such that it is an even function and is zero when integrated between the limits of ±∞. It is something.

Considering that p is a function of the variable t, and for all values of t, dp(t)/dt=q(t) q(r-h)=-d/dhp(r-h), If you write, becomes. q is an even function, and its integral converges as described above, and the limiting range (integral range) ±1/2 A is sufficient for f(h) to converge to zero within that range. Since it is large, the first term on the right side of the last equation is zero. accordingly , where the function p is an indefinite integral of the function q. Any integral p is the convolution value C(r)
It can be shown that it can be increased by any finite number without changing the value of . In general, apart from certain components, the function p is an odd function, and this property is illustrated in FIG.

The convolution integral in f(h) can be divided into a number of partial integrals according to the form of partial integrals described above so that this integral can be converted into a finite sum, and each term of the sum is , the sampled value of f'(h) multiplied by the corresponding convolution coefficient determined by the shape of the function p in the integral range of the problem. The sampled derivatives, which are the samples of the function f′(h), can be usefully represented by the finite difference values mentioned above, and they are generally arbitrary with respect to the equipment involved. It may be formed in any suitable manner, eg as described below with respect to FIG.

This diagram may be considered as relating to the determination of the absorption coefficient at one or more points within a planar region of the object to be examined. For simplicity, it is assumed that the distribution of absorption coefficients is circularly symmetric and that the coefficient at one point on the axis of symmetry needs to be determined. Its axis of symmetry is perpendicular to the plane of the figure, so the first
It is perpendicular to the Cartesian axes O _x and O _y shown in Figure 0. The lines marked L are parallel to the O _y- axis and are one of the parallel lines that are close to each other and equally spaced.
are representative of one set, along which the total transmission of radiation is observed in each case with a suitable measuring device. O For a line x apart from _{the y-} axis,
Assume that the output to be measured of the corresponding device is I(x). The symbol B indicates a circle bounded by a radius R from the origin of the coordinate Q, and the boundary is considered to represent the edge of the object to be inspected. Point p whose coordinates are x, y
Assume that the coefficient is f(r), in which case r ² =x ² +y ² .

In this assumed situation, it can be expressed as I(x)=∫ ^+Y _-Y f(r)dy. However, Y ² =R ² −x ² .

Since the following function rdr=rdy holds along the path of integration, the expression for I(x) can be rewritten as follows.

Ix=2∫ ^R _x f(r)rdr/(r ² −x ² ) ^1/2 In this equation, I(x) means the Abel transformation of f(r), and correspondingly, There are inverse transformations such as

f(r)=-1/x∫ ^R _r I′(x)dx/(x ² −r ²
) ^1/2 However, I'(x) = dI(x)/dx. Thus, the coefficient on the axis of symmetry is given by:

f(o)=-1/x∫ ^R _p I'(x)dx/x In finite series form, assuming that the spacing along the parallel paths of integration has a sufficiently small value a, this result becomes In the finite difference form, It is expressed as However, Δ(n+1/2・a)=I(+1・a)−I(na). The convolution of the derivative I'(n+1/2・a) or its corresponding finite difference Δ(n+1/2・a) is the convolution at the origin according to each sum with the function 1/(n+1/2). Gives the value of the absorption coefficient.
The restriction of a circularly symmetric distribution can be removed if a large number of parallel sets of observed data, represented by the function I(x), are first obtained,
These pairs have small equiangular spacing from 0 to 2π
It will be clear that they are arranged over a range of inclinations up to. Integrating the convolution sum over this angular range, the asymmetric component of the distribution to be determined undergoes cancellation in the asymmetric pair, and the required value given by the integrated convolution of the symmetric component of the distribution Leave the decision.
Therefore, the process steps described above can be applied without any symmetry restrictions.

It will be appreciated that finite differences formed according to the techniques described above will have a small magnitude compared to the values they differ from. Thus, according to the invention, digital multiplication of convolutions can be performed in less time than in the processes contemplated in the above-mentioned specification. Special request 1977-
Although the apparatus described in Japanese Patent Publication No. 47032 (Japanese Patent Publication No. 59-11150) can be used to perform this convolution, the only change seen by using the difference signal rather than the net signal is It is called the L coefficient in the above specification.

Although the present invention has been described above in relation to an apparatus that inspects only a single planar virtual cross section of an object to be inspected, it is possible to appropriately overlap the detectors and their related circuits, and By changing the shape of the X-ray beam emitted from the cross section, two or more planar virtual slices can be examined simultaneously.

The detector need not be constructed with individual crystals and secondary electron multipliers. In a modified configuration, a single large detector crystal is used, with different regions of it coupled to each photodetector.

Furthermore, although in the above-described apparatus the angle of the radiation beam 23 was sufficient to encompass the entirety of the object in the plane of interest, this need not be the case; in the modified configuration: The tube 9 is adapted to generate a fan beam of smaller angle and is scanned directly at the radiation source and its corresponding detector relative to the object to be examined, in order to allow proper inspection of the object to be examined. Exercise is given. This linear scanning motion is in addition to the rotational scanning motion of the radiation source and detector relative to the object to be inspected, and the two scanning motions are synchronized.

[Brief explanation of the drawing]

1a is a side elevational view, partially in section, of a device according to the invention; FIG. 1b is an arrow bb in FIG. 1a;
2 is a schematic diagram of a device of the type shown in FIGS. 1a and 1b together with the circuitry that makes predictions during operation of the device; FIG. A schematic diagram illustrating the principle of operation of the apparatus shown in the figure, Figure 4 is similar to Figures 1 and 2.
5 is a cross-sectional view of the X-ray tube used in the apparatus shown in the figure, FIG. 5 is a view showing details of the tube shown in FIG. 4, and FIG. FIG. 7 is a block diagram showing the computational circuitry for generating the output display from the apparatus shown in FIG. 2; FIGS. 8 and 9 illustrate one aspect of the invention; FIG. The graph, FIG. 10, shows a processing method based on using a difference signal for a net signal.

Claims (1)

[Scope of Claims] 1. In an apparatus for inspecting a substantially planar cross section of an object to be inspected using penetrating radiation including X-rays, the apparatus comprises: a radiation source configured to emit a radiation source; a plurality of detectors configured to detect radiation after passing through the object along a plurality of paths at a plurality of different angles within the fan; and a common radiation source that intersects the cross section. a device for rotating at least the radiation source relative to the detector about an axis; and a device for rotating at least the radiation source relative to the detector about an axis; A radiographic inspection device equipped with a device that rotates periodically.