GB1577046A

GB1577046A - Radiographic scanner apparatus

Info

Publication number: GB1577046A
Application number: GB12048/77A
Authority: GB
Original assignee: Ohio Nuclear Inc
Current assignee: Ohio Nuclear Inc
Priority date: 1976-06-01
Filing date: 1977-03-22
Publication date: 1980-10-15
Also published as: CA1096512A

Description

(54) RADIOGRAPHIC SCANNER APPARATUS (71) We, OHIO-NUCLEAR INC, a corporation organised under the laws of the State of Ohio, United States of America, of 6000 Cochran Road, Solon, Ohio 44139, United States of America, do hereby declare the invention, for which we pray that a patent may be granted to us, and the method by which it is to be performed, to be particularly described in and by the following statement: This invention relates to radiographic scanner apparatus.

Computerised tomographic X-ray scanners are known, for example see U.S.

Patents Nos. 3,778,614 (Hounsfield) and 3,924,129 (Le May). Previous data handling systems were basically software systems and were not as fast as desired. As a result, it was necessary to allow the data to build up in a computer before the image could be viewed on a display unit. As a result, if there were errors in the scanning operations, for example due to patient movement, a large portion of the test was conducted before the error could be noted. The net result would be that the complete test would have to be repeated.

In order to increase the speed of the data handling system in order to produce an image for viewing, much of the data handling in a preferred embodiment of this invention described below is accomplished by hardware. That is, the preferred embodiment of this invention includes a hardware system or processing means which operates faster than software. Moreover, because many of the software functions have been removed from the computer, the computers that are necessary in preferred embodiments need only have a reduced capacity and are, therefore, smaller and less expensive.

According to a first aspect of the present invention there is provided a radiographic scanner apparatus for measuring the intensity of radiation after passage through a planar region and for reconstructing a representation of the attenuation of radiation in the planar region by a medium disposed therein, the apparatus comprising a source of radiation for irradiating the planar region along sets of paths spanning the planar region; detector means for detecting radiation traversing the planar region along the set of paths and for producing a set of data output values, each output value being indicative of the intensity of radiation along one of the paths of the set of paths, the set of output values forming a data line, the detector means being disposed generally opposite the planar region from said source; means for rotating at least the source of radiation for irradiating the planar region along successive sets of paths disposed at a succession of angular orientations about the planar region, whereby the detector means is operative to produce a succession of data lines corresponding to the succession of angular orientations; a convolver means for convolving each of the succession of data lines with a filter function to produce a succession of convolved data lines, the convolver means comprising a data memory for storing each of the succession of data lines, a filter memory for storing a set of filter values which comprise the filter function, means for multiplying each data value of a data line stored in the data memory by each filter value stored in the filter memory to produce a plurality of products, and product accumulating means for forming a set of sums by summing preselected groups of the plurality of products, said set of sums constituting said convolved data line; and means for processing the convolved data lines to create the representation of the radiation attenuation in the planar region.

According to a second aspect of the present invention there is provided a radiographic scanner apparatus comprising means to generate a succession of data lines each comprising a set of data values, each data value being indicative of radiation intensity or attenuation along a determinable path, and a convolver for convolving each of the succession of data lines with a filter function to produce a convolved data line, the filter function comprising a set of filter values, and the convolver comprising: a data memory for storing each of the succession of data lines; a filter memory for storing the filter function; means for multiplying each data value of the data line stored in the data memory for each filter value stored in the filter memory to produce a plurality of products; and product accumulator means for forming a set of sums by summing preselected ones of the plurality of products, said set of sums constituting said convolved data line.

The invention will now be further described, by way of example, with reference to the accompanying drawings, in which: Figure 1 is a block diagram of a radiographic scanner apparatus or system embodying the invention; Figure 2 is a perspective view of a scanner and patient table of the apparatus; Figure 3 is a schematic representation of motion of a radiation source and detector of the apparatus; Figure 4 is a schematic representation of a collimated X-ray beam; Figure 5 is a graphical representation of a filter function and raw data; Figure 6 is a graphical representation of convolved data; Figure 7 is a block diagram of a convolver of the apparatus; Figure 8 is a graphical representation of a convolution function; Figures 9 to 12 are schematic representations of a back projection imaging process; Figs. 13 and 14 are block diagrams of an imager system; and Fig. 15 is a block diagram of scaling hardware.

The radiographic scanner apparatus or systems illustrated in Fig. 1 comprises an Xray scanner 10, a convolver 12 and an imager 14. Computers are used in the implementation of the various functions.

The scanner 10 has included with it circuitry including a log amplifier 16 which receives data 18 from the X-ray scanner 10. A reference signal 20 is passed through a similar log amplifier 22 and compared with the data signal in an amplifier 24. The resultant signal is converted by an analogto-digital converter 26 and operated on by a computer 28 in order to make software corrections in the signal. The resulting corrected signals are sequentially passed into the convolver 12. The convolver 12 operates on the resultant data in accordance with a filter function 29 and creates, for example, a 256-word convolved data line.

The number of words may, of course, be varied. The resultant information is passed to an image-storage disc 30 and from there through a computer 31 to the imager 14 with a feedback 36. A video display 38 permits viewing of the output. In general terms, the X-ray scanner generates data which needs to be processed before it can be imaged.

The convolver 12 is a hardware processor of the data. It performs a function which was previously done with computer software.

Because the convolver 12 works with hardware, it provides the necessary alteration of the basic raw data at a much faster speed than otherwise would be normally possible in a computer with software. The convolved data is then used to create an image by means of imaging circuitry which is also accomplished by hardware.

The general imaging technique as illustrated in the above-mentioned U.S.

Patent No. 3,778,614 (Hounsfield), the disclosure of which is incorporated by reference. While the technique is generally the same, the convolver and imager of the present apparatus or system vary greatly from the techniques described therein. In general, the imaging technique employed by this system utilizes a collimated radiation beam, normally of X radiation, and a detector array to make a series of transverse scans across a patient at various angles throughout 1800 of axial rotation. Beam attenuation data is measured at intervals during each scan and is mathematically processed to form an image.

Fig. 2 illustrates the physical embodiment 40 which houses the X-ray scanner, and it includes a source of radiation shown in phantom as 42 which transmits radiation, usually X radiation, to a detector 44 on the other side of an opening 46 in which a patient or other medium to be subjected to the radiation would lie. The source of radiation 42 and the detector 44 are mounted for a linear traverse and rotational movement at the end of each traverse. It is anticipated that various types of sources, for example fan beam, and different movements, for example purely rotafional, could be used.

The radiation source 42 and detector 44 along with some appropriate circuitry and framework are mounted for traverse and rotational movement in a casing 48 having a front face 50, top 52 and side 54. Mounted at the sides of the casing 48 are vertical support members 56 and 58 which support the casing 48 with its internal mechanism for rotation on shafts (not shown). The vertical members 56 and 58 are supported on a horizontal member 60 and connected thereto.

A patient support means 62 has a table 64 with rails 65 and 66 supported on a vertical stand 67 with a front face 68 and side face 70. Elongated rail-like members 72 and 74 add stability as a base for the vertical support 67.

The source 42 and detector 44 make a large number of traverses and rotational movements. It is necessary to view 3600 of the slice under examination, but a 1800 rotation is normally sufficient to provide that data. The 1800 rotation may be done in various manners. In practice, 60 rotational movements of 3 each may be used or 180 movements of 10. By way of example, Fig. 3 illustrates the traverse movements of the source of radiation 42 and detector 44 at angles of 45 , 90" and 135". A square target 82 is used to illustrate the problems involved with the scan procedure. At each angle, raw data is generated and is shown diagramatically for the three angles as 76 at 45 , 78 at 90" and 80 at 1350. At 90 , the X radiation data gives a true representation of the absorption of the X-rays passing through the square target 82. However, at 450 and 135 , the X radiation passes through different thicknesses of the square target 82.

As a result, the raw data output would indicate a different shape than what is actually known for the target. The X .radiation received by the detector 44 bears an inverse proportion to the distance it travels through the target medium 82. It is this problem, in large part, which necessitates use of the present system.

One potential embodiment of the source of radiation and detector is illustrated in Fig. 4. It includes a source of radiation 84, such as a therapy-type X-ray tube, which is collimated by a collimator 86 having three rectangular apertures 88, 90 and 92.

Collimated beams 94, 96 and 98 are thereby formed and diverge at an angle of about 2" from each other. A second collimator 100 receives the beams of radiation and separates each beam into two portions by means of pairs of openings 102, 104 and 106 in order to further collimate the beams for detection by pairs of photomultiplier tubes 108, 110 and 112. The photomultiplier tubes work in conjunction with scintillation crystals which change the X radiation to visible radiation. Several types of commercially available scintillation crystals, such as NaI, are known to those skilled in the art.

The X-ray beams are collimated at both the source and the detector so that only narrow beams of X-ray photons are used for imaging. The output amplitude of each photomultiplier tube 108, 110 and 112 is proportional to the number of X-ray photons and their respective energy levels (beam intensity) which reach the scintillation crystals. The beam intensity, in turn, is inversely related to the density, atomic composition, and a total length of the material in the path of the X-ray beam.

That is, with all other factors constant, less photons reach the detector as material density increases since a greater number would be absorbed or scattered by the material. The ratio of X-ray beam intensity received at the detector with respect to the intensity at the source (IJIs) is defined by an exponential function of the linear attenuation coefficient (y) and respective lengths (x) of the various material along the length of the beam.

In the present system, the beam intensity (ID) measured by the detectors is converted to a logarithmic value by a logarithmic amplifier as noted earlier in Fig. 1. In addition, a reference detector measures the unattenuated beam intensity (l5) at the X-ray source. The reference level is also converted to a logarithmic value and is subtracted from the logarithmic value of the detected beam. The resultant value, which is entered into the system in digital form, is a measure of the total beam as follows: Mt=ln IDIn Is Thus, the beam attenuation value entered into the system increases as the beam intensity measured by the detectors decreases.

The beam attenuation values (8t) are sampled at various points as the X-ray beam and detector are moved across the target body during a traverse scan pass. The series of beam attenuation values at each of the sample points is entered into a system as a data line. The traverse scanning trajectory is then axially rotated to a new angle and another line of beam attenuation data is entered into the system. In practice, the beam path at each sample point for the new angle is projected through a different combination of tissue region and, therefore, will result in different beam attenuation values. It is this characteristic which permits a cross-sectional image representing a map of the linear attenuation coefficients within the scanned area to be reconstructed. As illustrated in Fig. 3, a crude image of the scanned area can be obtained by simply back projecting each raw data line onto an image matrix. That is, th measured data values are projected with respect to an image matrix. That is, the measured data data line was measured. Thus, each image matrix point accumulates the data value that was measured when the X-ray beam was projected through its corresponding point in the scanned area. The effect is as if the measured data values had been paint rollered across the image matrix.

A problem occurs because the attenuation profile, as illustrated in Fig. 3 of the final reconstructed image line, shows a hump-shaped profile rather than the sharp rectangular attenuation profile of the scanned object. This distortion partially results because each value that is back projected to a matrix location represents the attenuation contribution along the entire beam length. To compensate for this effect, the raw data values are combined with a preset set of additional values. The set of additional values may be called interchangeably the convolution function or filter function. The combination of the filter function and raw data generate a convolved data value which, in effect, weights the raw data values in accordance with its position relative to the filter function curve. Fig. 5 shows, at the top portion thereof, a graphical representation of the filter function, and at the bottom portion thereof the raw data from a 900 traverse. The filter function shape is such that all values along the raw data line which represent undesired contribution to the image reconstruction are made negative and are appropriately weighted in accordance with their respective contributions. A basic equation for the filter function is given by: -1 for for k=0, +1 +2...

(4K2-l) where F,k, is linear between integer values of k and where k equals the number of projection distances from the point being constructed.

When convolved data values are used, the undesired contribution from surrounding areas will be eliminated. In order to obtain a complete line of convolved data values for back projection as described in more detail hereinafter, the filter function is combined with the raw data in such a manner that the filter function is repositioned to obtain each convolved data value. For comparison, Fig.

6 shows the same reconstruction at 900 except that convolved data values are illustrated instead of raw data values. In this example, the attenuation profile of the reconstructed image line closely matches the attenuation profile of the scanned object.

The convolution function is carried out by what is termed herein as a convolver and that term is used to mean hardware means for combining the raw data with a filter function. The convolver executes in hardware the key mathematical operation used in the convolution algorithm for image reconstruction. In a continuous space, this operation is called a "convolution integral" and has the form:

This integral has the property of multiplying the frequency spectrum of F by the frequency spectrum of G yet always operating in the spatial or time domain. The convolver performs the discrete convolution integral, convolving a 256-word (16-bits per word) data array with a 256word filter function. It is understood that various capacities may be used and the one described herein is merely for the preferred embodiment. The filter function is a fixed array whose values are picked to have a particular frequency spectrum.

As illustrated in Fig. 7, the convolver has a computer input to a data memory 120 and a filter function memory 122. The outputs from the data memory 120 and filter memory 122 are entered into a 16-bitx 16-bit binary multiplier 124. The resultant product from the binary multiplier 124 is accumulated in a 40-bit product accumulator 126. A scale factor shift register 128 accepts the output from the accumulator 126 and passes it to an operatively connected convolved data storage buffer 130. In operation, all convolver functions are commanded by the software by intermediate circuitry. The software can command the convolver to input data to the data memory 120 or filter memory 122. One way of causing such transfer is through a non-processor (NPR transfers). Once an NPR data input or NPR filter input operation is initiated, the convoiver performs NPR transfers from the addresses defined by the convolver address system register until the memory is filled. In this case, 256 words will fill the memory. The convolver then notifies the software that the operation is completed via a processor interrupt. The filter memory 122 is then commanded to initiate a series of 256 convolution cycles which calculates and transfers a complete line of 256 convolved data values into the system. During a convolution cycle, each of the 256 raw attenuation data values in the data memory 120 is multiplied by its corresponding filter function from the filter memory 122. An example of this multiplication is illustrated in Fig. 8, in which A and B represent the first two data points and C the last data point in a traverse.

In the graphical representation of Fig. 8A the point CDo is found by making the calculation as follows: CDo=Fo . D0+F1 . D1+F2 . D2 . . . +F255 . D255 After a convolution cycle is completed, Fig.

8B shows that the filter function is, in effect, shifted so that the new data point would be calculated as follows: CD1=F, . D0+F0. D1 +F, . D2+ . . .+F254 . D255 After the second convolution cycle, as after the first, the convolved data value is calculated and stored. The calculation continues until the last point is calculated as: CD255=F255 . Do+F-254 . p1+ F253 . D2+ . . . +Fo . D256 After the convolver has calculated and transferred a complete line of 256 convolved data values into the system for image construction, the convolver requests a processor interrupt. The interrupt informs the software that the operation is finished and the convolver is ready to accept raw attenuation data for the next line.

Each resultant product from each convolution cycle is accumulated (algebraically summed) in the 40-bit product accumulator 126. After all 256 products have been calculated and summed, the convolved data value from the accumulator 126 is "scaled" or given a range.

Output scaling is accomplished by selecting 16 bits from the 40-bit product accumulator 126 which will be stored for output as a convolved data value. The scaling factor is established by a software program. The 16-bit convolved data value is entered into the convolved data storage buffer 130 to complete a convolution cycle.

Although the convolved data storage buffer 130 can accommodate up to 64, 16-bit words, the convolver would normally -initiate a request for transfer as soon as data is available in the buffer for output.

As noted earlier, the convolver described herein has substantial advantages over performing the calculations in a computer as was done previously. The computer takes a relatively large time measured in at least several minutes to compute the convolutions. On the other hand, the described convolver can do 180 convolutions in about nine seconds. The advantages of this substantially reduced time permits real time imaging of one slice during the scanned period. A second slice or traverse pass through the subject takes only about fifteen seconds to process, thus, not contributing to system dead time between scans. Utilization of hardware for the convolution process also permits the use of a much slower and less expensive computer.

Because the filter function is perfectly symmetrical, it is only necessary to store half of it in the memory if so desired. Logic is subsequently changed so that equivalent computations are made.

The imager (14 Figure 1) performs the second step in the imaging procedure. It accomplishes the back projection operation and provides a video display of 256x256 image memory matrix. There are several ancillary functions which are incorporated by the imager, but the one of interest herein is the "scaling" of the display. Scaling, as used herein, means viewing a limited range of densities in the target area.

Back projection acts to locate where a projection from an image matrix location falls on an input data line. As seen in Fig. 9, a scan center is located between the source of X radiation 42 and detector 44. The scan motion is at an angle 0 from a predetermined reference direction 134. In order to accurately reproduce the target scanned, a similar image center 136 is established in a 256x256 image matrix 138.

The identical angle 0 is established between the image matrix and a back prqjection input line as particularly noted in Fig. 10.

While it is theoretically possible to back project by taking each point in the input data line and tracing it through the image matrix, the problem arises that the projection misses most of the memory locations. In a strict mathematical sense, the projection of a point never strikes a matrix point.

The back prqjection process and apparatus described herein moves point by point through the image memory and calculates where a projection from an image matrix location falls on the input data line.

For example, in Fig. 11, the image matrix point 140 projects onto the input data line, perpendicular thereto, between the data points 142 and 144. The value to be added to 140 must be linearly interpolated from the values at point 142(a) and point 144(b).

Linear interpolation is an averaging process, weighted by the relative distances from the projection of 140 to 142 and 144. Generally, if the projection of point 140 lies XO/, (percentage of a unit spacing data line) from 142, the value added to the image matrix point 140 will be:

It is important, therefore, while determining these projections from image matrix points onto the input data line, to keep track of both an integer address (address of points on input data line) and a fractional address (projection's spacing between points on input data line). This fractional address (the "X" from the equation above) will determine the weighting factors for the linear interpolation.

To enable the back projection system to compute the "projection address" (location of the projection from a given matrix point onto the input data line) for any point in the image, three numbers are computed and loaded into the imager by the processor.

Since the back projection operation starts (for convenience only) at the upper left corner of the image matrix, the projection of this point (address of X=0 and Y=0, or 0,0) onto the data line must be provided as a number called IZP (Image Zero Point).

Referring to Fig. 12, the computer computes IZP (0,0) from the Image Center address on the input data line and the angle 0. Calling Rot the image center value: IZP=Rot-127.5 COS0+127.5 SINO The back projection starts at point 0,0 (where its projection equals IZP) and "steps" point by point through the matrix in a raster scan fashion. That is, it steps along X axis (horizontally) until the end of the line (X=255) then steps down one line starting at the left side again. The computer supplies two values to permit this operation: ISX, which is the distance on the input data line corresponding to 1 unit of X (horizontal) of the image matrix, and ISY, which is the distance on the input data line corresponding to 1 unit of Y (vertical) of the input matrix. These values are: ISX=COS0 ISY=-SINe Thus, the projection of any X, Y point in the image matrix is: Proj (X,Y)=IZP+X (ISX)+Y (ISY) The system to calculate Prqj (X,Y) relies on the fact that the back projection operation scans through the image matrix by incrementing X (sequential horizontal points) to the end of the line (X=255). The incrementing Y (sequential lines) and repeating to the end of the matrix (X=255, Y=255). So rather than doing the multiplication implicit in the Proj (X,Y) equation, the Projection Address hardware can do successive additions.

A system shown in Fig. 13 for accomplishing the back projection includes holding registers 146, 148 and 150 for ISX, ISY and IZP, respectively. The respective values for ISX, ISY and IZP are fed into the registers by computer. Two control lines from a reset X sum 151 and clock X sum 153 are fed into and control an X accumulator 152. The reset X sum sets the accumulator 152 output to zero. The clock X sum will cause ISX to be added to the current sum.

The same system is used for a Y accumulator 162.

The X accumulator 152 includes an X sum latch 154 which continuously adds the data in the accumulator to the new data by means of an adder 156. Similarly, a clock for the Y sum 158 and a reset Y sum 160 feeds the Y accumulator 162 which includes a Y summation latch 164 feeding back to an adder 166. The output of the Y sum latch 164 is fed into an adder 168 and its output added to the X sum latch 154 by an adder 170 and transmitted to projection address.

When starting the back projection operation of image matrix 0,0, both accumulators 152, 162 will be reset and, therefore, the projection address will equal IZP. To determine the projection address for the second point, the X accumulator 152 will be clocked and the projection address will equal IZP+ISZ. For each successive point across the first X line at Y=0, the X accumulator 152 will be clocked so that the nth point the projection address will equal IZP+N . ISX. At the end of the X line, where Y=0, that is after the 256th point, the X accumulator will be reset (X=0) and the Y accumulator clocked (Y+l). The process is then repeated and continued for all 256 Y lines.

As noted in Fig. 14, the projection address consists of both an integer (whole number) and a fractional component. The integer component looks up the appropriate input data value and the fractional component determines the interpolation weight values, which are complementary fractions that add to unity. That is, for each image matrix point, a two-step process through an adder 172 and a two's complement circuit 173 is necessary to compute the value to be added to the current memory value. First, the integer address's data value times (I-fraction), i.e.

unity minus the fraction, is loaded into an accumulator 171. Then, the address data value for the integer plus unity times the fraction, is added to the accumulator. The sum is then added to the image matrix point for which the projection address was calculated. The projection memory stores 256 input data points, each as 16-bit words.

However, since the multiplier circuitry employed cannot multiply negative numbers, the negative numbers are converted to positive numbers by a negative number adjuster 174 with a sign bit at the data input to the projection data memory 175. After the multiplication by multiplier 176, if the data value had a negative sign b multiplications by iterative adding of the numbers in the accumulator. The accumulator 171 receives reset and clock signals at the appropriate times. An adder 180 combines signals from the accumulator 171 and the memory output data to form memory input data. The imager may process multiple image matrix points simultaneously by the use of additional circuits illustrated in Fig. 13 and 14.

As noted above, the imager further includes the system for scaling which concentrates on certain densities of target subject matter. Fig. 15 illustrates the system for scaling. Utilization of circuitry or hardware instead of software for this function allows for instantaneous changes in a range of data being displayed, thus, allowing interactive (no time delay) adjustment of the range of differential numbers being displayed.

An image memory 200 transmits data to a 20-bit subtract data offset 202. Also feeding the 20-bit data offset 202 is a 20-bit offset holding register 204. This offset corresponds to the lowest level of the densities of material desired to be displayed on the video display. Since the image memory data does not correspond one-to-one with the differential numbers involved, generally a density number change is represented by 10 to 20 number change in the image data value. The computer calculates the appropriate offset value and enters it into the offset holding register 204. If the offset subtraction generates a negative result, the output value will be forced to zero. This corresponds to attempting to display an image point lying below the bottom of the number range selected.

The offset data value is applied to a digital gain control 206. The digital gain control 206 has a two-stage gain control. The first stage gives binary steps of gain (2, 4, 8, 16, etc.). The second stage is a multiplier which multiplies the output of the first stage by numbers ranging from 1.00 to 1.995. The offset establishes the appropriate differential number representing radiation levels for the bottom of the displayed range. A 12-bit gain holding register 208, signaled from a computer 210 sets the span or range of the displayed rang. Image memory data values falling below the bottom of the range will be negative values. Data values offset falling above the top of the range will be over-scale values. Negative values force the output of the gain control circuit to adopt the value 0.

Over-sclae values force the output to adopt its maximum possible value 63.

The output of the gain control, a number ranging from 0 to 63, looks up an 8-bit video intensity value from a display look-up memory 212. This memory is loaded from the computer with 64 values of 8 bits each to program the presentation of the data. In practice, there may be four such memories if the output is to be in color. An 8-bit output of the display look-up memory 212 is applied to a digital-to-analog converter 214 which produces a voltage signal to drive a video display 216.

Again, the use of hardware to accomplish the functions set forth above give the distinct advantage in time to the operation of the X-ray scanner. Previously, many of the functions carried out in such devices were done in software.

Dynamic imaging, as used herein, generally means the visual representation on a video output of the reconstructed image of the medium very quickly after the detection of the X radiation. Dynamic imaging also operates in conjunction with the imager as used herein. In particular, at the end of each traverse scan, software terminates the traverse, and initiates rotation to the next angular position.

When the imager is running, the computer 31 scans the image status block looking for a complete set of data. The imager will back project the data and the computer will store the back projected data on discs if available. If the computer 31 does not find many complete sets of data needing to be imaged, rather if a scan is in progress, it will discover a partial set of data and back project whatever data it has. The block diagram of this is shown in Fig. 1. The computer 31 will continue to monitor the status of the image and back project new data as it becomes available, doing so until the image is completed. When the image is completed, it is stored on disc 30 and the computer 31 continues to search for a new data.

By utilizing a system which allows an image storage disc 30 to be used for the video display at any time, it is possible to visualize on the video display 38 in Fig. 1 the data as it is being gathered or within a second or two thereof. Because of this dynamic imaging approach, it is possible to note any problems with the test as it is being conducted. That is, if a patient should move and the image is not being reconstructed correctly, the test can be stopped and restarted at that time without the necessity of waiting until the test is at a much later stage.

This invention has been described with reference to a preferred embodiment with some possible modifications thereto.

Obviously, other modification and alterations will be obvious to others upon the reading and understanding of this specification. It is our intention to include all such modifications and alterations within the scope of the appended claims.

Claims

WHAT WE CLAIM IS:

1. A radiographic scanner apparatus for measuring the intensity of radiation after passage through a planar region and for reconstructing a representation of the attenuation of radiation in the planar region by a medium disposed therein, the apparatus comprising a source of radiation for irradiating the planar region along sets of paths spanning the planar region; detector means for detecting radiation traversing the planar region along the set of paths and for producing a set of data output values, each output value being indicative of the intensity of radiation along one of the paths of the set of paths, the set of output values forming a data line, the detector means being disposed generally opposite the planar region from said source; means for rotating at least the source of radiation for irradiating the planar region along successive sets of paths disposed at a succession of angular orientations about the planar region, whereby the detector means is operative to produce a succession of data lines corresponding to the succession of angular orientations, a convolver means for convolving each of the succession of data lines with a filter function to produce a succession of convolved data lines, the convolver means comprising a data memory for storing each of the succession of data lines, a filter memory for storing a set of filter values which comprise the filter function, means for multiplying each data value of a data line stored in the data memory by each filter value stored in the filter memory to produce a plurality of products, and product accumulating means for forming a set of sums by summing preselected groups of the plurality of products, said set of sums constituting said convolved data line; and means for processing the convolved data lines to create the representation of the radiation attenuation in the planar region

2. Apparatus according to claim 1 wherein the convolving means further includes a scale factor shift register and a convolved data storage buffer operatively connected to each other to receive the sums from the product accumulator means.

3. Apparatus according to any one of the preceding claims wherein the multiplying means has a 16-bitx 16-bit capacity, and the product accumulator means has a 40-bit capacity.

4. Apparatus according to any one of the preceding claims wherein the filter memory stores the filter function defined by: -1 Fk= for k=0, +1 +2 4K2-l where Rk is linear between integer values of k and where k equals the number of projection distances from the point being constructed.

5. Apparatus according to any one of the preceding claims wherein the data lines and filter functions are combined according to the formula:

6. Apparatus according to any one of the preceding claims wherein the means for processing further includes an imager, the imager comprising means for back projecting convolved data lines on an image matrix.

7. Apparatus according to claim 6 wherein the back projecting means includes holding registers for ISX and ISY, where ISX is the distance on a convolved data line corresponding to one unit in the X direction in the image matrix, and ISY is the distance on a convolved data line corresponding to one unit in the Y direction in the image matrix.

8. Apparatus according to claim 7 wherein the back projecting means also includes a holding register for IZP, where IZP represents a corner of the image matrix.

9. Apparatus according to claim 8 wherein the imager further includes an X accumulator and a Y accumulator operatively attached to the ISX holding register and the ISY holding register, respectively.

10. Apparatus according to claim 9 wherein the imager further includes a first adder for adding the output of the IZP holding register and the Y accumulators; and a second adder for adding the outputs of the X accumulator and the first adder to form a projection address which address may consist of an integer component and a fractional compdnent.

11. Apparatus according to claim 10 wherein the imager further includes a third adder for adding zero and positive or negative integer values to the integer component, a two's complement circuit addressed by the fractional component to produce the complement of the fractional component, a convolved data line memory operatively connected to receive the address from the third adder, and a multiplier for multiplying convolved data values from the convolved data line memory by the fractional components and the complements of fractional components.

12. Apparatus according to claim 11 wherein the imager further includes a first negative number adjuster for changing negative sums to positive sums in the convolved data lines stored in the convolved data line memory and a second negative number adjuster for converting any negative number back to positive numbers at the output of the multiplier.

13. Apparatus according to claim 12 wherein the imager further includes an accumulator for iteratively adding the output of the negative number adjuster.

14. Apparatus according to claim 7 wherein the means for processing further includes a means for scaling the display including an image memory and a holding register having their outputs directed into a subtract data offset means for subtracting an offset value received from the output of the holding register from the output of the image memory.

15. Apparatus according to claim 14 wherein the means for scaling the display further includes a gain control connected to receive the output from the subtract data offset, the gain control including a binary gain factor and an interpolation multiplication.

16. Apparatus according to claim 15 wherein the means for scaling the display further includes a display look-up memory arranged to receiver the signals from a selector.

17. Apparatus according to claim 16 wherein the means for scaling further includes a digital-to-analog converter to change the signal from the display look-up memory from digital-to-analog form for transmittal to a video display.

18. Apparatus according to claim 6 which further comprises dynamic imaging means comprising an image storage disc having an input operatively attached to the convolving means and an output operatively attached to the imager with a feedback therefrom so that an image may be displayed on a video output during the accumulation of data.

19. A radiographic scanner apparatus comprising means to generate a succession of data lines each comprising a set of data values, each data value being indicative of radiation intensity or attenuation along a determinable path, and a convolver for convolving each of the succession of data lines with a filter function to produce a convolved data line, the filter function comprising a set of filter values, and the convolver comprising: a data memory for storing each of the succession of data lines; a filter memory for storing the filter function; means for multiplying each data value of the data line stored in the data memory by each filter value stored in the filter memory to produce a plurality of products; and product accumulator means for forming a set of sums by summing preselected ones of the plurality of products, said set of sums constituting said convolved data line.

20. Apparatus according to claim 19 wherein the multiplying means is a binary multiplier.

21. Apparatus according to claim 19 further comprising a scale factor shift register and a convolver data storage buffer operatively connected with the product accumulator means.

22. Apparatus according to any one of claims 19 to 21 wherein the filter memory stores the filter function defined by -l Fk= fork=0 +1, +2 4K2-1 where Fk is linear between integer values of k and where k equals the number of projection distances from the point being constructed.

23. A radiographic scanner apparatus substantially as herein described with reference to the accompanying drawings.