FR2582425A1 - Method and device for generating a tomographic image of a three-dimensional object - Google Patents

Abstract

The invention relates to a method and device for generating a tomographic image of a three-dimensional object according to a sectional plane P. This method is of the type consisting in irradiating the object at different incidences by beams converging towards the sectional plane P, and, for each incidence, in detecting the radiation after it has passed through the object, with a view to generating an electrical image signal. According to the invention, the various image signals thus generated are processed by comparing their values at each point and by selecting, for each image point, the value of the signal which corresponds to the lowest attenuation of the radiation; the values selected supply a synthesising signal representing the tomographic image. The method of the invention makes it possible to obtain a tomographic image which is strictly free of fuzzy traces originating from points situated outside the sectional plane P.

Description

(1_) FRENCH REPUBLIC Publication No. 2,582,425 (use only

  for INSTITUTE NATIONAL Reproduction Orders)

OF THE PROPERTY = INDUSTRIAL TE

  PROPERTY OF THE NATIONAL REGISTRY N 0 85 08066 PARIS

  (Int Cl ': G 06 F 15/68, A 61 B 6/03, G 01 N 23/08.

@ APPLICATION FOR PATENT OF INVENTION A1

  Filing Date: May 21, 1985. (Applicant (s): OKSMAN ESTABLISHMENTS

limited liability. - FR.

() Priority:

Q Inventor (s): Edgard Garcia.

    Date of the public availability of the

  application: BOPI c Patents "n 48 of November 28, 1986.

  References to other national documents appearing

paid: t) Holder (s)

  Agent (s): Cabinet Barre-Gatti-Laforgue.

  Method and device for generating a tomographic image of a three-dimensional object

  The invention relates to a method and device for general

  tomographic image of a three-dimensional object along a section plane P. This method is of the type consisting of

  irradiate the object under different impacts by means of

  C beams converging towards the cutting plane P and to detect for each incidence the radiation after passing through the object to generate an electrical image signal. According to the invention, the different image signals thus generated are processed in

  comparing their values in each point and selecting.

  for each image point, the value of the signal corresponding to the attenuation of the weakest radiation; the selected I / values provide a representative synthesis signal L of the tomographic image. The method of the invention allows -. - - I

  to obtain a tomographic image rigorously free; -.

  I fuzzy traces from the points outside the plane of section P. N 0 <Lt IN Oee en fascicules' NATIONAL PRIMERIE. 27 rue de la Conventio75732 PARIS CEDEX 15 O DENT The fascicles at the NATIONAL IMPFIMERIE. 27. rue de la Convention -75732 PARIS CEDEX 15

  METHOD AND DEVICE FOR GENERATING AN IMAGE

  TOMOCRAPHIC OF A TPIDII1ENSIONAL OBJECT

  The invention relates to a method for generating a tomographic image of a three-dimensional object according to a cutting plane; it extends to a tomography device

  perfected for the implementation of this method.

  The tomographic technique is now well known to specialists and aims to provide an image

  representative of the section of an object by a specific plane.

  This technique is essentially applied in the medical field for the internal observation of tissues at certain levels. Its advantages over simple radiology techniques lie in obtaining a more meaningful image of the area to be observed, with reduction of defects from other areas. This technique consists in irradiating the object under different incidences, generating for each incidence a corresponding image signal, and processing these different signals to obtain a synthetic image; in a traditional way, this treatment consists of a photographic integration in order to reproduce a clear image of the points situated in the plane of crossing of the various incident beams. More recently (for fifteen years), have been developed special processes that achieve the integration electrically on electrical image signals (example: radiological tomography method described in FR No. 1,591,807

of May 4, 1970 in the name of SEREL).

  The defect of these processes, whether photographically or electrically integrated, lies in the complexity of the image and in its low densitometric resolution, which makes it difficult to interpret; indeed, if this image does include the net representation of the points of the interesting section plane, it also contains superimposed curves more or less fuzzy from other areas and this, because of the principle of integration that consists of adding all the

information collected.

  Some devices have been designed to try to remove this defect. The device described in French Patent No. 2,415,400 of January 10, 1979 in the name of PHILIPS comprises several radiation sources, which make it possible to obtain several images on a fixed receiver; these images are processed by an optical system, then by computing means to reconstruct a synthesis image at a chosen level. It seems naked the influence of the fuzziness can be diminished by this process but not rigorously suppressed; moreover, this technique is not adaptable to conventional tomographs and requires the realization of a special multi-source tomograph with a specific optical system, which explains why this technique is not implemented at all.

the current time.

  Another process described in French Patent No. 2,469,739 of November 5, 1980 in the name of PHILIPS consists in producing two photographic images on a tomograph whose source is subjected to two scans of different geometry (for example circular and linear); two images are then analyzed by two video cameras in order to eliminate the shadows of different geometries, which correspond to points outside the section plane.The implementation of this method is long and complex since it requires the taking and development two snapshots and further processing

by a separate device.

  Furthermore, in other types of equipment, it is known to perform image signal processing, adapted to the requirements of each application concerned. Thus, US Pat. No. 4,437,161 (ANDERSON) discloses an angiography device in which, on the one hand, a plurality of time-shifted images of an organ of a single and fixed radiation source are produced from a single radiation source. a patient after injection of contrast products, and on the other hand, a difference image is developed by subtracting from each image a stored mask constituted by one of the appropriately selected images; in the device of this patent, this difference image is compared pixel by pixel, at two fixed levels, in order to determine the validity of the mask used and to carry out, if necessary, the storage of the significant images received; the comparison serves only to eliminate the non-significant images (patient motion or the still untriggered action of the contrast product) but in no case does it cause a change in the received images. Significant images are then stored and processed by subtraction in a traditional way.

  angiography in order to obtain a synthetic image.

  However, these specific treatments do not solve the problem of tomographic blur and do not provide any education to those skilled in the art for

to devise a solution.

  The present invention proposes to solve this problem by a very simple and inexpensive implementation method, suitable for application on tomographs

traditional.

  The objective of the invention is therefore to provide, from a conventional tomograph, a tomographic image rigorously free of fuzzy traces.

  from points outside the section plane.

  Another object of the invention is to apply the method to the most widespread type of topograph comprising a mobile radiation source, a fixed object support to be examined and, opposite, a mobile transducer.

  according to a synchronous movement of that of the source.

  On the other hand, in some tomographs, the various positions of the radiation source (s) during the tomographic scan correspond to very different distances and beam incidences from the photosensitive surface of the transducer; these variations in distance and in incidence cause significant disturbances of the image since there then intervene non-neoligeable factors of attenuation of radiation, foreign to the object to be examined. No known device is able to overcome

of these defects.

  Another important object of the invention is to eliminate the influence of variations in radiation intensities on the image, due to variations in distance and

  incidence of the source relative to the transducer.

  The method according to the invention for generating a tomographic image of an object according to a cutting plane comprises the following conventional steps: (a) using at least one radiation source, irradiating the object under impacts different by beams converging towards the section plane, (b) by means of a transducer having a photosensitive surface, detecting for each incidence the radiation after passing through the object in order to generate for each of said bearings a signal representative of an image, called an image signal, (c) and processing the different image signals corresponding to the different incidences in order to deliver a

  synthesis signal, representative of the topographic image.

  According to the present invention, the processing (c) of the different image signals consists: (c) comparing the value of these different signals at each image point, (c2) to be selected for each image point the value of signal which corresponds to the attenuation of the weakest radiation, (c3) to memorize the selected values constituting the synthesis signal representative of the tomographic image, (c4) and to deliver said memorized signal to visualization means in order to view the tomographic image. As will be better understood later, such a method radically removes any shadow that has been displaced on the photosensitive surface during the different irradiations under different incidences. Indeed, for each point, is selected the value corresponding to the lowest attenuation for these different incidences, so the strongest illumination; if a shadow passes temporarily during one or more incidences on the point considered, it is not taken into account since the point keeps its value of maximum illumination; on the contrary, if a shadow is stable, the gray level of the point of the image will take account of it. Now stable shadows represent the points of the plane of section, while shadows that move represent points outside this plane. Consequently, by a very simple treatment, a clear image of the points of the cutting plane is preserved, free of insignificant fuzziness. This image thus gains in both spatial and densitometric resolution. The processing can be implemented directly on the detected image signals. In this case: (c) the value of the image signals (c2) is compared directly with the detection value, and the value corresponding to the radiation intensity is selected for each image point.

detected maximum.

  Generally, the detection (b) of the radiation generates for each incidence an image signal whose value is in each point a direct function of the intensity of the radiation received. It is then sufficient to select (c2) for each image point the maximum value of the various

  image signals. The simplicity of such a treatment is conceivable.

  According to another implementation mode intended to compensate for the influence of the variations of distance and angle of incidence of the source with respect to the photosensitive surface: (cO) the image signals are corrected before their comparison ( ) by modifying their values at each point of initiation, as a function of the variations of distance and angle of incidence of the irradiation, (c) the different corrected image signals are compared at each point of image, (c2) the maximum corrected value is

  selected at each image point.

  This embodiment provides an image representative of the cutting plane particularly clearly due to the removal of defects inherent in the displacement of the source relative to the object; this correction proved to be crucial for some very

widespread tomographs.

  If necessary, the implementation device can be equipped with a selector for shunt

or not the correction process.

  The treatment of the invention can be applied to any conventional tomograph, because it does not require any architectural modification of it: it is sufficient to equip the scanner with image acquisition means (generally already provided) and adapted processing means . Thus, the invention extends to an improved tomography device of the conventional type comprising a tomographic stand carrying a mobile radiation source, an ohjet support to be examined and a moving transducer in synchronous movement with that of the source, said transducer being provided with a photosensitive surface and adapted to deliver, from the received radiation, electrical image signals for generating an image on viewing means; according to the present invention, this device is equipped with: - an analog / digital converter arranged to sequentially receive the image signals coming from the transducer and adapted to deliver to each sequence a digital image signal, - storage means arranged to receive on their input at each sequence the digital image signal from the converter, said means being provided with a write command and adapted to store the value of selected pixels according to the state of its write command - Comparison means arranged to receive each sequence, on the one hand, a signal corresponding to the digital image signal from the converter, on the other hand, the signal contained in the memory means, said means of comparison being adapted to provide at their output, for each pair of pixels of the same address, a comparison bit depending on the respective value of these pixels, the output of the comparing means being connected to the write command of the storage means for generating or inhibiting, pixel by pixel, the storage of the value of the pixels present at the input of said storage means, - d a pilot oscillator and an address generator adapted to control the converter and the

storage means.

  Other features, goals and benefits

  of the invention will emerge from the description which follows in

  reference to the accompanying drawings, which illustrate an embodiment of the device according to the invention and an embodiment of the method; in these drawings: - Figure 1 is a schematic view of a conventional tomography device, equipped with processing means according to the invention, - Figures 2a, 2h, 2c, 2d, 2e, 2f are diagrams illustrating the concept implemented in the method of the invention, - Figure 3 is a block diagram of the processing means of the device of the invention, - Figure 4 is a diagram illustrating an embodiment of the method by comparison and successive selections; FIG. 5 is a symbolic diagram illustrating the correction process; FIGS. 6-a, 7a, 8a, 9 are detailed diagrams of subsets of the device, allowing operation in several parallel ways, FIGS. 6b, 7b, 8b are time diagrams, corresponding respectively to FIGS. 6a, 7a and 8a; FIG. 10 is a partial diagram of one embodiment of the device, FIG. time, illustrating the FIG. 12 is a partial diagram of a multiple sampling embodiment, FIG. 13 is an explanatory diagram of the embodiment of FIG.

  result obtained by this multiple sampling.

  The tomograph shown diagrammatically by way of example in FIG. 1 is of the traditional type. It comprises a tomographic stand equipped with an arm 1, pivoting about a hinge point 2 and carrying: at one of its ends, a source of radiographic radiation 3 (X-ray tube), at the end opposed, a transducer (4, 7) having a photosensitive surface, constituted in the example by the photosensitive surface of a gloss amplifier 4 adapted to radiology, at its point of articulation, a

support 5 for the patient.

  The support 5 is fixed during the tomography scan and can be adjusted beforehand in height relative to the articulation point 2, by a mechanical system

classic symbolized in 6.

  The radiation source 3 is permanently oriented towards the photosensitive surface of the image intensifier 4; the latter remains parallel to a given plane, generally horizontal, and constantly centered on the axis of the beam from the source. The scanning of the source generated by the rotation of the arm 1 can be performed according to different trajectories depending on the tomograph concerned (the photosensitive surface moving in synchronism with the source). In FIG. 1, an example of circular scanning has been drawn; during this scan, the source can be

  activated either continuously or pulsed.

  During scanning, the successive beams irradiating the object have different incidences and all converge towards a sectional plane P (the axis of the beams all intersecting at a point C of the plane of section, it should be noted that this "plane "may be a non-planar surface depending on the shape of the photosensitive surface, but this usual expression of" plane of cut "will be retained, being

  of course it must be interpreted broadly).

  We will not provide more detail on the general architecture of this type of tomograph, which is

  traditional (tomograph type "Polytome" PHILIPS).

  The image intensifier 4 is coupled to a video camera 7 which delivers video image signals corresponding to the image transmitted by the amplifier. In the case of a pulsed radiation regime, the camera outputs a video image signal for each different incidence of radiation firing. In the case of a continuous radiation regime, the camera delivers from the evolutionary image transmitted on the amplifier 4, a video image succession at a frequency that is a function of its acquisition time: it is considered in this case that the images obtained (in fact non-static) correspond to the effects of discrete radiation sampled at

acquisition frequency.

  The successive video image signals corresponding to the various incidences are delivered to processing means 8 including a production node

  preferential will be described later.

  These processing means 8 receive a position signal generated by means 9 for measuring the relative positions of the radiation source 3 and the

  transducer (4, 7). These conventional measuring means in themselves

  same (which are already provided on some tomographs), are associated with the arm 1 and provide position signals of abscissa and ordinate, a function of the angular position of this arm in space (these signals thus define the relative positions source and transducer). The measuring means 9 may be of the double potentiometer type actuated by a "joystick" secured to a

  deformable parallelogram mounted on the arm 1.

  The processing means deliver a synthesis signal to a video monitor 10 for viewing

  the corresponding computer image.

  As will be detailed below, the processing means 8 are adapted to compare, point by point, the value of the various image signals (corrected or not) and to select for each point the maximum value and store this value in order to to constitute the signal of

  synthesis delivered to the monitor.

  The diagrams of FIGS. 2a and 2h to 2e

  illustrate this process and the result obtained.

  Two absorbent points A and R of the object to be examined are considered, one A located in the plane of section, the other B located outside of it: these points are irradiated under different angles (three implications

  are shown schematically in Figure 2a: R1, R2, R3).

  For the radiation R1, an image signal 1n symbolized in FIG. 2h is obtained, in which the projections A and B provide two separate shadows, functions of the respective attenuTIONS of the points A and B (it has been assumed that the point B located in outside the cutting plane absorbed

  more and therefore gave a more marked shade).

  For radiation R2, points A and B give shadows that coincide (the attenuation being a

  direct function of the sum of attenuations (Figure 2c)).

  For the R3 radiation, the shadow of point B has shifted further, while that of point A has retained the same position on the photosensitive surface of

  the image intensifier (Figure 2d).

  The treatment which consists in selecting at each point the maximum illumination level (the lowest attenuation of the radiation) thus leads: for the image point (or pixel) corresponding to B1, to keep the maximum level of the three signals of image, so to remove the shadow B1 2n. for the pixel corresponding to A1, A2-82 or A3, to keep the maximum level of the three image signals, therefore, to keep the level of A1 or A3, for the pixel corresponding to B3, to keep the maximum level of three image signals, so to delete

the shadow [3.

  Thus, the synthesis image signal illustrated in FIG. 2e only includes the shadow of the point A located in the sectional plane with a truly significant gray level: the moving shadows of the point B situated outside the plane of In the example, these ombhers being more intense than those of the point of the sectional plane, the integrated image obtained in a conventional way would be of quasi-impossible interpretation.

  symbolized in broken lines in Figure 2f).

  FIG. 3 represents the block diagram of one embodiment of the processing means 8 targeted by the invention. The analog video signal delivered from the camera 7 is sent to an analog / digital converter 11; this converter controlled by a 1l pilot oscillator 12 is of the "Flash Converter" type, able to sequentially receive the emptyé inode signals and to deliver to each sequence a corresponding digital image signal: the analog video signal coming from the camera is sampled at the frequency of the pilot oscillator and the converter defines discrete image points (pixels) and generates digital image signals, each formed by a succession of words, representative of the succession of pixel values. Each digital image signal from the converter 11 is transmitted directly or in a corrected form, via a selector 13, to storage means 14 constituted by a random access memory equipped with a write command W and having a capacity adapted to memorize the different words constituting a digital image signal. In a conventional manner, the memory 14 is managed by an address generator 15 controlled by

the pilot oscillator 12.

  This generator 15 outputs address bits in a predetermined sequence, which are used to control a camera synchronization circuit 16, generating synchronization signals, horizontal and vertical, compatible with the video camera used to control the scanning of it in synchronism with the address generator; these sync signals can also be used to control the monitor scan

of television 10.

  Moreover, comparison means 17 whose inputs are connected to the input and the output of the memory 14, receive at each sequence, on the one hand, the signal corresponding to the digital image signal (image signal digital itself or corrected digital image signal),

  on the other hand, the signal contained in the memory 14.

  These comparison means 17 perform on these signals a comparison, pixel by pixel, and are adapted to provide at their output, for each pair of pixels of the same address, a comparison bit depending on the respective value of these pixels. This comparison bit is delivered through control logic 18 to the write command W of the memory 14. The logic 18 validates the writing order of the comparison means 17 at the appropriate times of each sequence and includes initialization means for erasing the contents of the memory 14 prior to each tomographic sequence. The comparison bits are elaborated by the comparison means 17 so as to: write a write to the memory 14 when the value of the pixel presented at the input of the memory is greater than the value of the pixel of the same address which, stored in the memory 14 is present at the output of this memory, to insert any writing in the opposite case. Thus, the comparison means 17 and the random access memory 14 perform real-time comparisons and successive selections relating to the different nuneric signals from the converter: comparison of the first two image signals, generation and storage of an intermediate signal of synthesis by selecting, for each image point, the value of the one of the two signals which corresponds to the weakest attenuation, 25. comparing this intermediate synthesis signal with another image signal to generate by a similar selection a new intermediate synthesis signal, and renewal of these operations between the new synthesis intermediate signal and another image signal and this, for all the image signals, the comparison being made at each step between the image signal concerned and the intermediate signal of synthesis

previous comparison.

  FIG. 4 illustrates the comparisons and successive selections in the case of the three radiations of FIG. 2a; the two by two comparisons of the signals

  lead to the final svnthesis signal already commented.

  The output of the RAM 14 is furthermore connected to the television monitor 10 via an adaptation interface, an example of which will be described by way of example below; at the end of the tomographic scan, the final computer image is displayed on this monitor. The selector 13 has in the example three input buses (FIG. 3), one connected directly to the converter 11, the other connected to a correction multiplier 19, and the third connected to a mass memory 20 which, on the if necessary, allows to work in deferred time. The selector can be manually or programmatically controlled, in order to select one of three operating modes: real-time operation without correction, real-time operation with correction, operation

in deferred time.

  The multiplier 19 receives at its inputs, on the one hand, the digital image signal coming from the converter 11, on the other hand, correction coefficients

  stored in a memory 21 for storing tables T1, T2 ...

  correction coefficients; the calculation and the loading of the correction coefficients in the memory 21 will be described

further. The selection of the correction coefficient table is carried out at

  each sequence (corresponding to a given incidence) by means of an interface disposed between the measuring means 9 of the tomograph and the storage memory 21. This interface comprises means 22 for converting analog / digital position signals from the means of measurement 9 and an addressing logic 23 adapted to deliver to the storage memory 21 addresses function of the measured positions. In addition, this logic allows the interior of each table to choose the correction coefficient corresponding to the point concerned, by providing the address of this coefficient in the table concerned.

function of the pixel concerned.

  The multiplier 19 receiving for each pixel the correction coefficient selected in the storage memory, operates the multiplication of the value of the pixel from the converter 11 by this coefficient and delivers the

result * to selector 13.

  The photosensitive surface of the transducer 4 is fictitiously divided into a plurality of elementary zones Z1, Z2, Z3 ... as symbolized in FIG. 5. Each correction table T1, T2 ... corresponds to a given incidence R1, R2. .. and contains a matrix of correction coefficients in number equal to the number of elementary zones Z1, Z2, Z3 ... All the pixels of the same elementary zone will therefore be affected by the same correction coefficient. During tomographic scanning, the table is chosen according to the incidence of radiation in progress, thanks to the coordinates provided by the measuring means. Then, the coefficient of the table is selected in it directly from

  the address of the pixel being processed.

  The zones Z1, Z2, Z3 ... being all identical, this addressing can be very simply carried out by retaining in the complete address of the pixel, the part

  address corresponding to the elementary area concerned.

  The correction coefficients are determined so as to compensate for the variations in intensity of the detected radiation, due to the variations of distance and angle of incidence of the irradiation. These coefficients can be calculated theoretically for each incidence and each elementary zone by considering the distance and the average incidence for the zone considered. However, it is simpler and more realistic to carry out a preliminary calibration consisting in: irradiating the photosensitive surface 4 in the absence of an object under various incidences, 30. storing the values of the detected image signals, in particular in the auxiliary buffer of a computer (not shown), and to calculate from these values the correction coefficient tables, each coefficient being a function inverse to the average value of an image signal

  on the corresponding elementary area.

  This pre-calibration is done once and for all on the site, once the device is in place; the computer can be a unit integrated in the device or

  an auxiliary computer connected to it.

  The calculation can be performed by taking as a reference value the average value of the signal in the central zone of the photosensitive surface when the radiation source is in axial position vertically to the center of the photosensitive surface; coefficient 1 is assigned to this reference value. The other coefficients are then constituted by the inverse of the mean value measured in the absence of objects in each elementary zone and this, what

  whatever the impact (ie for all tables).

  In the case of a continuous radiation or a pulsed radiation with a large number of incidences, it is interesting to carry out the preliminary calibration by selecting a limited number of incidence among all the incidences, by irradiating the surface in the absence of an object under these selected incidences, by memorizing the values of the detected image signals, by calculating the correction coefficients corresponding to these signals (in accordance with the process described above) and by extrapolating these calculations for the other effects in order to obtain the corresponding correction coefficients. This allows to establish and memorize the missing tables

  for unselected incidences.

  For example, in the case of a source with a circular path, it is possible for the calibration to select only two diametrically opposite incidences, the tables corresponding to the other incidences are

  deducing by a matrix rotation calculation.

  Moreover, the response times of the various constituent sets of the processing means (converter, multiplier, selector, memory, comparator, control logic, etc.) accumulate and may not be compatible with sampling frequencies.

high of the image signal.

  To solve this problem, demultiplexing means are arranged at the output of the converter 11 to create -n- parallel processing channels; the memory 14 is then divided into n subsets and the comparison means 17 into n comparators, each associated with a control logic 18 and assigned to a subset of memory. These demultiplexing means comprise, in the example shown in FIGS. 6a, 7a, 8a, 9: 5, means for generating clock and control signals, for each processing channel, means symbolized in FIG. 6a, means demultiplexing the data symbolized in Figure 7a, in. means for addressing the n subsets of the random access memory ia, symbolized in FIG. 8a, and data multiplexing means, symhholized in FIG. 9 in order to reconstitute the signal of

final synthesis.

  Figures 6-9 are for the case of two treatment pathways (treatment in a higher number of pathways

  can be done on the same principle).

  The pilot oscillator 12 supplies a signal CK0 at the base frequency controlling the converter 11 (signal

symbolized in Figure 6b).

  A divider 24 associated with an auxiliary logic 25 (FIG. 6a) generates: two complementary signals CK1 and CK2 of divided frequency (in the example of the figures, by a factor of 2), and auxiliary signals CKAUXi and CKAUx2 constituted by positive pulses at the end of the cycles

respectively of CK1 and CK2.

  The complementary signals CK1 and CK2 will constitute the clock signals for each processing channel; the signals CKAUX and CKAUX2 will validate the write command in each subset of the memory

live 14.

  Of course, in case of a number of channels -n-

  greater than 2, the frequency will be divided by this number n, the divider 24 and its auxiliary logic 25 being adapted to this

number of lanes.

  The denultiplexing means schematized as examples in FIG. 7a include -n- D-type registers such as

  26, arranged in parallel with the output of the converter 11.

  These registers store on the rising edge clock signals of the channel concerned CK1, CK2, the present value

on their entry.

  Figure 7b is a diagram of the times in

the case of two ways.

  Figure 8a shows the structure of the address generator working in multiplexed channels. It comprises 2 groups of counters such as 27 and 28 generating: 10. directly the addressing of a subset of RAM, and the addressing of the subset of the other channel via a register D type 29 synchronized

  by the signal of the corresponding channel CK2.

  Figure 8b is a dianramme of the times of

  this generation of multiplexed channel addresses.

  FIG. 9 represents, at the output of the sub-

  sets of RAM 14, the multiplexing means 30 for the reconstitution of the final synthesis signal; these means are constituted by a multiplexer connected to the outputs of the various memory subassemblies 14 and controlled, in the

  ceas of the 2-way example by the signals CKI or CK2.

  This multiplexer drives the television monitor 10 through a display interface comprising: a D type register 31 controlled by the oscillator 12 in order to obtain a signal that is not altered by the response time of the multiplexer, and a digital / analog converter 32 receiving data and timing signals to reconstruct a composite video signal adapted to the monitor

of television 10.

  FIG. 10 represents an embodiment of the comparator assembly 17, logic 18 and

  memory 14 (box in broken lines in Figure 3).

  This structure is found for all pathways

  treatment in the case where one works in several ways.

  At the output of the selector 13, the data are re-calibrated in time by a D-type register 33, in order to synchronize the arrival of these data with that of the addresses, themselves similarly readjusted at the output of FIG.

  address generator 15 by registers 34.

  At the output of the memory 14, a D-type register 35 controlled by the clock signal via an inverter 36 presents the memory output data on one of the inputs

of the comparator 17.

  The control logic 18 composed of an inverting gate, an AND gate and a gate OiJ receives a

  external erase command for initialization.

  This external command is also connected to a reset input of the register 33. This command, which can be manual, thus forces the writing in all the random access memory 14 of data reset to zero contained in the register 33. OR gate control by a signal synchronized with the clock of the channel concerned validates the comparison signal at an appropriate end of cycle time (o the

  comparison is significant as shown in Figure 11).

  This figure 11 is a diagram of the times

  relating to the embodiment of Figure 10.

  On the other hand, Fig. 12 partially shows an embodiment in which the input analog video signal is sampled at a frequency which is a multiple of the frequency of the digital image signal in order to

  achieve a more faithful digital conversion.

  In this case the oscillator 12 is adapted to deliver a frequency sampling sampling signal multiple of the base frequency; the converter 11 is itself adapted to respond to this higher frequency. This converter is outputted to temporary storage registers such as 37 arranged in parallel and controlled in a manner analogous to that of the registers 26 of FIG. 7a by means of a divider associated with the pilot oscillator (divider analogous to the divider 24 of the

Figure 6a).

  The registers 37 are connected at the output to an adder 38; the output of this adder is connected to a buffer memory register 39, controlled by the

divisor above mentioned.

  The words of the digital image signal present at the output of the register 39 constitute a digital representation of better fidelity than in the case of a

  sampled conversion at the base frequency.

  Figure 13 is a diagram that illustrates this

  result in the case of double frequency sampling.

  The first signal symbolized in this figure is an example of an analog video signal with frequencies close to the base frequency. The second signal symbolizes the digital signal sampled at the base frequency. We see that the first undulation is not restored. The third signal symbolizes the digital signal sampled at twice the base frequency. The average value of value pairs

  sampled accounts for both undulations.

Claims (14)

  1 / - Method for generating a tomographic image of a three-dimensional object according to a cutting plane (P), consisting of: (a) using at least one radiation source (3), to be irradiated the object under different incidences (R1, R2 ...) by beams converging towards the cutting plane (P), (h) by means of a transducer (4, 7) having a photosensitive surface, detecting for each incidence the radiation after passing through the object in order to generate for each of said incidences an electrical signal representative of an image, called an image signal, (c) and to process the various image signals corresponding to the different incidences in order to deliver a synthesis signal, representative of the tomographic image, said method being characterized in that the processing (c) of the different image signals consists: (cl) to compare the value of these different signals by each image point, (c2) to be selected for each image point the value of u signal which corresponds to the attenuation of the weakest radiation, (c3) to memorize the selected values constituting the synthesis signal representative of the tomographic image, (c4) and to deliver said memorized signal to viewing means (10) ) to view the image tomographic.   2 / - Method according to claim 1, characterized in that: (c) the value of the image signals (c2) is compared directly after detection (b) and for each image point the value corresponding to the maximum detected radiation intensity. 3 / - The method of claim 2, wherein the detection (b) of the radiation is carried out for each incidence so as to generate an image signal whose value is in each point a direct function of the radiation intensity received, characterized in that (c2) the maximum value of the different image signals is selected for each image point. 4 / A method according to claim 1, characterized in that: (cO) the image signals are corrected before their comparison (cl) by modifying their values at each image point, so as to compensate for the variations in intensity of the image. radiation detected, due to variations in distance and angle of incidence of the irradiation, (c) the various corrected image signals are compared at each image point, (c2) the maximum corrected value is selected in each image point.   / - Method according to claim 4, characterized in that the correction (cO) of the image signals is carried out: (d) by first dividing the photosensitive surface of the transducer (4, 7) into a plurality of elementary zones (Z, Z2, Z3 ...), (e) performing a precalibration consisting of: (el) irradiating at different incidences the detection surface in the absence of an object, (e2) storing the values of the signals of detected images, (e3) and calculating tables of correction coefficients (T1, T2 ...) each corresponding to an incidence (R1, R2 ...), each correction coefficient being a function inverse to the average value of an image signal on one of the elementary zones, (f) by multiplying in each image point the value of each detected image signal corresponding to a determined incidence, by the correction coefficient contained in the corresponding table, coefficient concerning the area   in which is the image point considered.   6 / - Method according to claim 5, characterized in that the preliminary calibration (e) is carried out: (eO) by selecting a limited number of incidences among all the incidences, (el) by irradiating the detection surface in the absence of object under these selected incidences, (e2) by storing the values of the detected image signals, X10 (e3) and calculating a number of tables corresponding to the total number of incidences by means of a   extrapolation for unselected incidences.   7 / - Method according to one of claims 1, 2,   3, 4, 5 or 6, characterized in that the comparison (c1) of the different image signals and the value selection (c2) are performed in real time by successive comparisons and selections consisting of: (g) comparing the two first image signals, (h) generating and storing an intermediate synthesis signal by selecting, for each image point, the value of the one of the two signals which corresponds to the lowest attenuation, (i) comparing this intermediate synthesis signal with another image signal to generate by a similar selection a new intermediate synthesis signal, (j) and to repeat these operations between the new synthesis intermediate signal and another image signal and this, for all the image signals, the comparison (i) being performed at each stage between the image signal concerned and the intermediate signal of synthesis resulting from from the previous comparison.   The method of claim 7, wherein: (b) the detection is associated with a frequency-selective sampled analog-to-digital conversion to define the image points (pixel) for generating digital image signals , each formed by a succession of words, representative of the succession of pixel values, (i) the two by two comparison of the image signals and intermediate synthesis signals is carried out, pixel by pixel, so as to preserve the value of the pixel corresponding to   attenuation of the weakest radiation.   9 / - Method according to one of the claims   preceding, wherein: (a) the irradiation is carried out by means of a single radiation source (3) set in motion opposite the cutting plane so as to provide successive beams having axes of different incidences (R1 , R2) all intersecting at a point (C) of the cutting plane (P), (b) the detection is effected by means of a transducer (4) moved in synchronism with the radiation source so that the photosensitive surface of this one   is constantly centered on the beam.   / - Improved device for generating a tomographic image with a view to implementing the method   according to one of the preceding claims, of the type   comprising a tonographic stand (1) carrying a moving radiation source (3), an object support (5) to be examined and a transducer (garlic, 7) movable in a synchronous movement with that of the source, said transducer being provided with a photosensitive surface and adapted to deliver, from the received radiation, electrical image signals in order to generate an image on display means (10), said device being characterized in that it comprises: an analog / digital converter (11) arranged to sequentially receive the image signals from the transducer and adapted to deliver to each sequence a digital image signal, - storage means (14) arranged to receive on their input at each sequence the digital image signal from the converter, said means being provided with a write command (W) and adapted to store the value of selected pixels according to the state of its control riture, - means of comparison (17) arranged to receive each sequence, on the one hand, a signal corresponding to the digital image signal from the converter, on the other hand, the signal contained in the storage means, said comparison means being adapted to provide at their output, for each pair of pixels of the same address, a comparison bit according to the respective value of these pixels, the output of the comparison means being connected to the write command (W) storage means for generating or inhibiting, pixel by pixel, storing the value of the pixels present at the input of said storage means, - a pilot oscillator (12) and an address generator (15), adapted to control the converter (11)   and the storage means (14).   11 / - Device according to claim la, characterized in that it comprises: - measuring means (9) relative positions of the radiation source (3) and the transducer (4, 7) adapted to deliver a signal of position, - a memory (21) for storing correction coefficient tables, - an interface (22, 23), arranged between the measuring means and the storage memory, and adapted to deliver to the latter addresses depending on the positions measured, - a multiplier (19) arranged to receive at each sequence, on the one hand, the digital image signal from the converter (il1), on the other hand, the selected content of the storage memory of the tables (21 ), said multiplier being connected to the comparison means (17) to provide these   last a corrected digital image signal.   12 / - Device according to claim 11, characterized in that the comparison means (17) are connected by an input to the converter (11) and the multiplier (19) through a selector (13) adapted to direct to said comparison means either the digital image signal or the corrected digital image signal.   13 / - Device according to one of claims 10,   I1l or 12, characterized in that the comparison means (17) are adapted to supply at their output a write control bit 48 when the value of the pixel resulting from the storage means (14) is less than the value of the other pixel   from the converter (11) or the multiplier (19).   14 / - Device according to one of claims 10,   11, 12 or 13, characterized in that the above-mentioned storage means (14) consist of a random access memory, of suitable structure and capacity, so that they can be stored in memory.   words constituting a digital image signal.   / - Device according to one of claims 10,   11, 12, 13 or 14, characterized in that demultiplexing means (2429) are arranged at the output of the converter (11) to create parallel processing paths.   16 / - Device according to one of claims 10,   11, 12, 13, 14 or 15, characterized in that: - the pilot oscillator (12) is adapted to provide control of sampling of the converter (11) at a frequency that is a multiple of the frequency of the digital image signal the converter (11) is connected in output to temporary storage registers (37) arranged in parallel, the pilot oscillator (12) is associated with a divider to ensure the successive control of said storage registers (37) to the frequency of the digital image signal, - said storage registers (37) are outputted to an adder (38), said adder (38) is outputted to a buffer memory register (39) controlled by the divider for successively outputting the words of the digital image signal.   17 / - Device according to one of claims 10,   11, 12, 13, 14, 15 or 16, characterized in that it comprises initialization means (18) adapted to erase the contents of the storage means (14) prior to each tomographic sequence.   18 / - Device according to one of the claims   in which the transducer comprises a radiology amplifier (4) adapted to radiology and a video camera (7) connected thereto for outputting a video signal, characterized in that the pilot oscillator (12) and the address generator (15) are arranged to control synchronization of the video camera to synchronize the video image signal with respect to said generator addresses (15).   19 / - Device according to claim 18, wherein: - the address generator (15) comprises counters (27, 28) connected to the pilot oscillator (12) for outputting address bits according to a predetermined sequence, - a camera synchronization circuit (16) is arranged to receive these address bits in order to deliver synchronization signals, horizontal and vertical, controlling   the scanning of the video camera (7). tt --- X ,, s3 '- 9 \\ NI Fig. 2b Fig. 2a RI _ R2 R3 (R1) A1 Fig. 2c A2 (R2) B2 r .. Fig. 2d A (R3)   A3 Fig.2e Sinl i Fig.2e J final synthesis Fig.2f [. ,, r., LJ LJI r 93 A3 A2 A1 B1   B2 13 l l I 19 l 14I Towards means -'14 1 d IL .__ S i - Visualization _ 'i, Oscittator 12 O To 11 -12 converter  ynchro. horiz.-1 nchro. horiz.   I122 TO Synchro. green. Synchro. green. CAMERA 7-   CAMX 1 [--- (1Two means of x | | 16 visualization Fig. 4 200 Fig.5 (Ri) 101lovW) o 3 to_200 300 _______ Zj (R 2) 4 <0 Intermidiary signal Incidence Ri Incidence R2 of synthesis 1 $> 1,2 1,12 1,05 1,2B 1,31,197   (Z1) (Z2) (Z3) (Z1) (Z2) (Z3)] (R3) 2- z. 1,17 4 0 A Synthetic C o signal 3 | fi nal to the synthesis s / 8 Fig. 6a CK1 12 i> Y $ - CK A & UXl 242 CK AUX 2 24, CK 2 CKO FIG. 6b C KO J I T! ' ! At z t t 'A; CK1 % '% I! I CK2 ,, I CK AUX2   Fig. 7a Video signal CK1 Oscillator 12 Fig. 7b t t + i t + 2 Osc 12 J1 F'LJ-1_ A Conv l CK 2 Channel 1 t t + 2 X Channel 2 t-1- t + 41 t + 3 y CKi F1 7i JL CK2 Fig. 8a 6J8 Synchro. horiz.   Mimoire 14 Memory 1% addresses horiz. addresses j / 1vertical lane lane 1 CK2l2 K Mimoire 14 Mimolarity addresses horiz vrsaes I // verticaLes flies 2 lanes 2 Fig. 8b t t + lt + 2 CKO CK-i CK2 -!, - '- 1 n--' - Voiei - + 2, J I Way2 t-: tJ-   Synchro Synchro Fig.9 green horizon ^ 32 l i CK1 or CK2 12 7/8 Fi g. 10 Order effaceme ._- CKAUX1 [ I Fig. He Addresses Addresses t i horiz. green.   Addresses es $, + ltX t +? 1 register - I Non-register output 35t + 1 t + 2 / I /   Sort ie / I compare invalid 1-7 CKAUX 1   Sointerite I Cycle t c ycle t + 2 8/8 CK1 F912 CK2   or CK2 Fig. 13 S / I l l video Ji 2 1 I   tI I Basic sampling Basic frequency sampling Double frequency sampling