FR2888105A1 - Patient`s body X-ray or infrared imaging method for scanner, involves making term to term averaging of adjusted and superimposed images to obtain synthetic image expressing coefficient of patient`s examined body - Google Patents

Abstract

The method involves averaging an acquired data in detectors (13a, 13b) for a pair of orthogonal rotation angles, into average values. An initial image is constructed with the average values, and an attenuation coefficient is adjusted by a method of least squares to obtain an adjusted image. The data averaging, the image construction and the coefficient adjustment are repeated for data acquired with different pairs of angles of rotation. A term to term averaging of adjusted and superimposed images is made to obtain a synthetic image expressing the coefficient of a patient`s examined body. Independent claims are also included for the following: (1) a device for implementing a method for X-ray or infrared imaging of a body (2) a computer duly programmed for implementing a method for X-ray or infrared imaging of a body (3) a computer program for implementing a method for X-ray or infrared imaging of a body or a device that implements a method for X-ray or infrared imaging of a body.

Description

X OR INFRARED IMAGING DEVICE

  The invention relates to an X or infrared imaging device of a body, more particularly comprising a support for receiving a body to be examined, a source emitting an X-ray beam or a light beam in a direction of propagation for irradiating or illuminating the body to be examined, an irradiated or beam-illuminated detector for detecting an attenuated intensity in consideration of X-ray or light-beam crossing through the body to be examined, and an analog-to-digital converter for converting the detected intensities into data enabling determine an attenuation by the body to examine X-rays or light.

  Such a device is known, for example, from US Pat. No. 3,924,131 or US Pat. No. 3,919,552. It will be recalled that CT (or CT scan) was discovered by an engineer of the EMI firm, GN Hounsfield, in 1968. The 1972 patent 'entitled "A method and apparatus for the examination of a body by radiation such as X or gamma-radiation". This invention earned the NOBEL prize in 1979 for its inventor.

  The principle is as follows: An X-ray beam, sweeps a defined plane, it crosses an organ linearly and strikes a plate or X-ray detector. Crossing the organ causes attenuation of the beam, the measurement can be made through the detector. Scanning crosswise in the cut plane produces a series of information processed by appropriate software on an associated computer.

  Indeed, following each scan axis in a heterogeneous medium the attenuation can be expressed by an exponential law, taking into account the photoelectric absorption and Compton scattering.

  Let 10 be the reference value x the value at a point X. We can write the following relation: 1 = loe-F In = IoE r Ln = O = fo A (x) dx From which we deduce by discretization: t Io _ + + + A Ln In = 0 A (x) dx - A, X, A2X2 X The successive values, AI, A2, An correspond to the values of each segment defined by X ,, X2, Xn.

  One can then express by a series of equations, the profiles of each scan associated with a precise angle (or position).

  A particular scale can be defined by the value, relative to a reference value of the attenuation coefficient, that of water, for example, or any other molecule suitably chosen.

  The scale used most often is that relating to a molecule, abundant in all living organisms, water.

  If we call A (h2o), the water attenuation coefficient, we can use a relative scale such that: Bn = [An - A (h2o)] * 1000 / A (h2o) The value of the coefficient water can be defined as 1 or 0, thus creating an easy-to-use scoring system since water is an essential component of the human body.

  Other systems can be used, depending on how the information obtained will be expressed (visually). The value of 1000 for the bone and 1000 for the air is often chosen.

  Computer processing of a sufficient number of crosswise scans, in fact defining small cells or elementary areas, allows the resolution of a set of linear equations provided that the number of scans is equal to the number of cells.

  The editing and use of the information is done by an associated computer.

  The computer collects all the data and thus calculates the value of the attenuation coefficient of each elementary zone.

  The information resulting from the calculations is translated into a map of the tomographic section plane.

  The set of cards constitutes the three-dimensional scanner image of the analysis, which allows longitudinal or transverse cuts.

  The medical interpretation is thus based on a true internal image of the tissues.

  Such images make it possible to check the state of certain bones, as well as the state of the brain, to detect a tumor or other abnormality.

  The explorations are preceded or completed by other explorations, for example ultrasonic ultrasound or magnetic resonance imaging.

  The scanner and the methods he has initiated remain an essential tool of medical exploration.

  Initially, a series of angular displacements of the order of 3 degrees was made, repeated a hundred times.

  The improvements made since allow to associate several beams with detection strips of a sufficient length to multiply the number of measurements made at the same time with multiple detectors.

  Fifth-generation scanners use detector strips perpendicular to the section planes to prevent movement.

  The image obtained results from a process in stages: - Obtaining the attenuation values for each projection, - Calculating the values of a profile, - Matrix representation of each section plane, - Translation of each representation by a particular map, - Establishment of a spatialized cartographic system.

  We arrive today at volumes of each elementary zone of the order of mm 3.

  However, it is far from the microscopic scale since the number of living cells is of the order of one billion in a mm 3.

  Early detection of cancer assumes a considerable gain in definition. But the time of use of the system for a given patient can not exceed an obvious economic threshold. But especially the increase in the number of profiles increases the dose of irradiation.

  It is admitted that a cancerous nodule accelerates its development when it causes an associated vascularization, this phenomenon occurs when a critical size is reached, for example 50 microns. In conventional processes, the irradiation dose and the calculation time are multiplied by 8000 to reach this level of fineness.

  The object of the invention is to modify a device known in the state of the art mentioned above to reduce both the irradiation dose and the computation time of the computer in the data processing for and so to increase the definition of exploration and resulting images.

  For this purpose, the subject of the invention is a device according to that referred to in the introduction, characterized in that it comprises - means for rotating the angle of rotation of the support mounted movably about an axis of rotation by with respect to the source and the detector mounted on a frame or for rotating the source and the detector mounted on a mobile frame about an axis of rotation with respect to the support and, a computer duly programmed to perform the following steps: (1) average the data from the conversion of the detected intensities, in n average values within n intervals resulting from a first partition of the data corresponding to a first division of the object to be examined into n elementary layers parallel to the direction of propagation of the beam for a first angle of rotation, and average the data from the conversion of the intensities detected, in m average values within m intervals resulting from a second partition data corresponding to a second division into m elementary layers parallel to the direction of propagation of the beam for a second angle of rotation, preferably different from 90 degrees of the first angle of rotation , the cut-outs in elementary layers making a grid in nxm elementary zones of a sectional plane of the object to be examined defined by the first and the second direction of propagation of the beam for respectively the first and the second angle of rotation, (2 construct an initial matrix (n, m) with the n and m mean values by assigning to each elemental area a row and column term (Bij) representing an attenuation coefficient defined by the sum of the average value on the interval (i) of the same line as the term, divided by the number (m) of columns of the initial matrix and the mean value over the interval the (j) of the same column as that of the term, divided by the number (n) of rows of the initial matrix, (3) adjust the attenuation coefficient in each elementary area by a least squares method taking into account constraints imposed by the border values of the average values using the following formula: Cij = Bij + n * (pj EBij) + m * (ci E; 1Bij) - nmJ (E1, Pj IijBij) where, in this formula, Cij = the desired value Bij = the estimated value initially (n) = the number of rows of the initial matrix (m) = the number of columns of the initial matrix n Ei -1Cij = pj for all the values of i, the stress of the column j E m Cij = ci for all the values of j, the constraint of the line i, J = 1 to arrive at a rectified matrix, (4) repeat steps (1) to (3) for data acquired with different pairs of rotation angles, and (5) average term futures the rectified matrices for different pairs of rotation angles to result in a synthesis matrix expressing an image of the attenuation coefficients of the examined body (9) under a definition determined by the grid.

  Preferably, the source has a broad emission tip, for example several centimeters in diameter for a cylindrical tip, emitting a beam also wide to irradiate or illuminate in a single control pulse of the source, a wide range of the body to examine and detector, - the computer being properly programmed to perform the following additional steps: (6) record the data from the conversion of detected intensities across the irradiated or illuminated range of the detector, (7) call the corresponding data to a particular sectional plane by selecting, from the recorded data, those resulting from the conversion of the intensities detected in a slice of the irradiated or illuminated range of the detectors, and (8) performing the steps (1) to (5) from these selected data.

  The first improvement concerns the acquisition of data, relating to the point attenuation coefficients. The author chose to use preferably the projection of the image on a detector. This method is normally intended to obtain an X-ray of the object to be examined, leading to producing a radiographic image on a computer screen and then on a printer.

  The acquisition of information can be done in two ways: by producing the radiographic image and then treating it, or by taking the information at the output of the detector, by processing it by converting the analog information into digital information and then processing this information.

  In this way, a considerable mass of information is available, since, as we shall see below, an existing detector on the market makes it possible to collect several millions of elementary information corresponding to pixels whose size is of the order of 25 microns. The multiplication of information obtained instantaneously by this technique leads to treatment needs themselves considerably amplified, which led to a review of the treatment process.

  The second improvement therefore concerns the processing of information by substituting for conventional linear algebra processing another method which can be summarized as follows: a) For example, in a first image obtained, a section or slice corresponding to a thickness is recorded. for example 1 mm and can go down to 25 pm, this slice being divided into elementary layers forming a grid of elementary zones of 1 mm 2 of 1 mm. Using the image reading software, it is possible to measure the average of the elementary intensities, to deduce the average of the absorption coefficients by the body to be examined, for each elementary square and to draw the attenuation coefficient associated with each square. . From each section, a first vector is deduced.

  Then we take a second image obtained at a different angle shifted for example 90 degrees by rotation of the object or shooting. A strip is cut in the same way in the same section plane from which a second vector is drawn. There are thus two orthogonal vectors.

  b) These two orthogonal vectors (or in any case secant), can make it possible to generate a first initial matrix in the following way: each vector is considered as a first series of the values of the edges of the matrix. Each row or column of the matrix is divided by a number equal to the number of terms of the other vector, which gives two sets of elementary terms associated with the different rows or columns. We can then evaluate each term of the matrix by taking the arithmetic (or geometric) mean of the term corresponding to the line and the term corresponding to the column on which the term is located. An initial matrix corresponding to a millimetric division has thus been created if the vectors themselves are thus obtained.

  c) This initial matrix is adjusted by considering the terms of each vector as constraints of line or columns, by the method of adjustment which is described below.

  This matrix of results can be translated by an image on a screen already giving a first image. To complete the search, we can realize a series of pairs of images offset by 90, from which we draw as many elementary matrices as couples. We can then calculate a matrix of terms equal to the average of the homologous terms of the individual matrices. We can finally calculate a standard deviation for each term and if the set is satisfactory produce the image corresponding to the matrix of means.

  The advantage of the device according to the invention is twofold: the acquisition of data is much faster and less restrictive than by the point-to-point scanning system. If the detector has a pixel definition of approximately 25 microns (for example the detector of the company ATMEL whose size is 23x6 cm offers on its surface 22 million zones of 25 microns of side), it allows to obtain several millions points even if free spaces are taken into account.

  By high resolution, a single flash for a few milliseconds replaces several millions of scans, resulting in the physical scanning being replaced by a series of computer scans.

  Thus the amount of radiation involved is particularly small.

  the data processing can be extremely simplified by calculating the adjustment thus used from the outset, then reiterated to obtain a number of adjusted matrices equal to p / 2 for p shots. These shots at regularly staggered angles during a complete turn are matched by staggered pairs of 90. The synthesis image is obtained by superposition in the same plane after rotation, and by calculating the averages of the homologous values obtained.

  For a matrix of 20,000 terms, the computation time is less than the second instead of being of the order of the hour on a PC computer available on the market.

  But it will be objected that an image obtained by a wide beam may be of lesser quality, in fact each ray of the slightly conical beam produces an image influenced by the images of neighboring rays or by echoes due to strong singularities (for example to metallic inclusions ).

  Experience shows that this influence is not considerable. In addition, a mathematical treatment can facilitate the analysis, in fact the values obtained on a row or a column of the matrix results can be treated as readjustable values by an adjustment for example polynomial, which has two consequences, the random errors are smoothed, but it is also possible to improve the definition by interpolation between the measured points and use of the recovery function taking into account two or more crossed interpolations.

  This research results in a very large economy of technical or computer resources in a field where the cost remains today too high to consider in many countries systematic explorations in the search for diseases such as cancer.

  The following description therefore exposes the actual calculation method of adjustment according to the invention. This method plays an important role in the processing of the signals resulting from the measurement, by the X-ray detectors, of the intensity or the residual value. of the elementary beam produced by the X-ray device, after the course in the body to be studied.

  If we want to treat a matrix of dimensions n rows and m columns if we call BU the estimated value at the line i and the column j, if we call Cij the most probable value of the corresponding term of the matrix, if we call pj the sum of the terms of the column j, if we call cj the sum of the terms of the line i.

  The estimate of Bij results either from the matrix amplification process, or from any other method allowing such an estimation, in particular from linear or polynomial fitting techniques.

  In this case, we will look for the solution of the Cij values taking into account the constraints of lines and columns, ie the minimum of the function: E (Cij Bij) 'For all values of i and j under the constraints: E Cij = pj for all the values of j Cij = ci for all the values of i The search for a minimum of the function under constraints will be made using the method of the multipliers of Lagrange, the Lagrangian will be written: L = - Eij (Cij Bij) 2 + E ,, (Ln =, (Cijpj)) + gi ((Cij This function is composed of two parts, a first does not have a 20 left character and the second is a set of linear relations.

  The Lagrangian is therefore differentiable for the variables Cij and j and fci, Lagrange multipliers associated with the constraints of rows and columns (we have in fact two groups of constraints, line constraints and column constraints).

  In these conditions we are able to obtain a set of linear relations concerning the Cij by derivation of the Lagrangian and a set of value of relations relating to the values of constraints, which is written: By specifying that the dL / dCij designates a partial derivative, of the function L for the variable Cij.

  dL = -2 (Cij Bij) + j + i = 0 1 dCij and the constraints Cij = pj, for all j Cij = ci, for all i 1 a Cij = Bij + X, j + i 10 2 The set of n * m relations corresponding to the partial derivatives plus the n + m relations of constraints is linear and admits only a solution corresponding to the nm + n + m variables.

  If for example we want to treat a matrix where n, the number of rows, is equal to 25 and m, the number of columns, is equal to 30, the solution by the linear algebra consists in processing: 750 variables Cij corresponding variables to the row multipliers, the pi variables corresponding to the column multipliers, the aj A first objective has already been reached since only 55 profiles must be established instead of 750.

  We have a total of 750 relationships corresponding to the partial derivatives and 55 relations corresponding to the constraints, for 805 variables. The resolution of this problem by using matrix calculation is the most obvious solution but it involves very heavy calculations, slightly heavier than those implied by conventional methods. The author initially relied on the improvement nr = 1 m Ei = i rapid computational processes, but beyond what was its primary goal, the radiation rate limitation being examined, it continued its research to try to improve the calculation time.

  From the combination of relations 1 and 2 we deduce: 5 En1Bij + 2 * + (En1, ui / 2) = p1 Im Bij + * pi + (> mj / 2) = ci 2 i = 1 We can deduce from these relations: xj = iJ * (2 * (pj-n1Bj) - ni) n 2 (ci -) - j) (m) pi = In these conditions, for example by referring the value of Xj in pi, we get: For all j = J * ((pj - En, Bij) - n, ui) n For all i pi _ W (Ein. = 1, ui) + 21 (ci-m thread

E *

  If we specify that p- = (1 / n) E (i = 1 to n) pi is the average of the multipliers bound to the constraint of the lines, we are led to the two following relations: For all j = 2 * (pj - En, Bi) - p - n For all i pi = p- +? * (ci - In 'Bij - mJ' - 'Indeed: m En' * In pi is equal to p - In these conditions by referring to the relation: Cij = Bij + ('/ 2) * (aj + pi) we are led to the algebraic relation. This adjustment formula makes it possible to deduce the matrix of Cij from the Bij matrix, by a term-to-term calculation C ij = B ij + \ n * (pj ElBij) + (* (ci E m BU) - fl * (1; 0 = 1 p - leu) The author has thus succeeded in quite surprisingly in a calculation of algebraic nature that does not require the use of matrix calculus.

  The algebraic method allows the partial processing of the reference matrix which in many cases may be sufficient.

  The numerical validation of this signal processing formula and establishment of the definition values of the medically sought image appears hereinafter.

  Example of application of the method, on a reduced model So be a matrix of n rows and m columns in which n = 3, m = 4

INITIAL MATRIX

  1 2 3 4 lines c 1 22 24 18 16 80 78 2 24 22 18 20 84 85 3 26 20 22 24 92 93 columns 72 66 58 60 256 P 70 67 59 60 256 In this matrix, the estimated values are entered on the three lines and the four columns, the constraints of lines are written in the column C. The constraints of columns are written in the last line P. The application of the formula above is simplified since the total of the constraints of columns (or of lines) is equal to the sum of the terms and leads to:

BALANCE AFTER CALCULATIONS

  1 2 3 4 E2 CA 1 20.83333 23.83333 17.8333 15.5 78 78 0 2 23.5833 22. 58333 18.5833 20.25 85 85 0 3 25.5833 20.5333 22.583 24.25 93 93 0 2 70 67 59 60 256 P 70 67 59 60 0 0 0 0 We can check , taking into account calculations made by a simple calculator that the value of the terms summed vertically or horizontally satisfy the constraints, but also that they lead to the desired results.

  Balancing using the linear algebra method This method expresses directly the linear relations between the variables Cij and Bij, and the variables Xj and pi.

  The classical method of solving a linear system involves the inversion of the matrix of the coefficients of the relations between the variables and the multiplication by this inverse matrix of the vector expressing the second members of the relations.

  a) There are 12 relations between the variables resulting from the expression of the partial derivatives of LAGRANGIEN of the form: Cij Xj / 2 - pi / 2 = Bij b) There are 4 relations of constraints relative to the columns and 3 relations of relative constraints to the lines, of the form: E: Ci = nCa = ci i = 1 J The classical method then requires the inversion of a matrix of a size equal to n * m + n + m is in our case 19x19, of which the calculation time is obviously much higher.

  To make it possible to verify the functioning of the device, the author has made a prototype comprising the following equipment: an X-ray beam, 70 kv, coming from a lead tube of a source 11 having a diameter of 6 cm and constitutes the tip 29, - a detector 13 of small size, 45 x 30 mm, - a means 17 for rotating an object to be examined 9 placed between the beam and the detector 13, with a step of 1 , (see FIGS. 1 and 2), a microcomputer 27 on which is housed medical imaging software (from KODAK), which makes it possible to obtain a radiographic image at different rotational steps of the object to be examined. 9, a) At first, 36 images were made with steps of 10, ie: 0, 10, 20, 90, 180, 270, 360 A 360 control image was thus made which is superimposed exactly on the image at 0, making it possible to verify that the rotation has been carried out correctly. By way of example, we append the following images: FIGS. 3a to 3d, the images at 0, 90, 180, 270 of the intensities detected.

  b) In a second step, a rectangular pair analysis was performed: - 0-90; 90-180; 180-270; 270-360 (0) c) Indeed, a certain distance from the reference plane (which can be seen in the appended images, a black spot which is the end of the rotating rod) is cut out , a narrow slice corresponds to a thickness of 1 mm, one thirtieth of the width of each image. Four bands corresponding to 0, 90, 180 and 270 are thus obtained.

  Each band is in turn cut into elementary squares corresponding in space to an elementary area of 1 mm 2, in the useful part of the image which is 30 mm (approximately). The average value of the intensity of the pixels in each of the 1 mm 2 areas is read with appropriate software. We obtain then a vector of thirty (or thirty-one) terms, corresponding to the averages of each of the elementary zones.

  As can be seen from Table 4, the values vary between 0 and 255. For 255, the elemental square is completely white; under these conditions it is assumed that the attenuation coefficient along the line leading to the corresponding term of the vector is equal to 255 minus the value of the luminous intensity of the homologous spot. Under these conditions, if there is no interposed object, the value of the attenuation coefficient is equal to 0. If the square is completely black, the value of the attenuation coefficient is equal to 255, the totality radiation from the beam was absorbed.

  The same treatment is repeated for the bands at 90, 180 and 270 which appears in Table 4, in the columns titled VAL CORR (corrected value). The corrected values appear for the four sections, 0, 90, 180 and 270. In this table, averages are estimated using appropriate software and standard deviations that remain within acceptable limits.

  Since the number of pixels recognized is high, the analysis can be continued down to a definition that is well below one millimeter. The values obtained are adjusted so that the sum of the values of each of the columns is equal to the average value, since under the same conditions of distance, the absorption of a given body is constant.

  d) Divide by 31 the previous term to deduce the value of horizontal (90) or vertical (0) constraint, these values appear in table 5. Table 5 (initial matrix) thus shows: the sums of the lines and columns, - the constraints drawn from table 4, - the differences between sums and stresses, - the deviations reduced by the division by 31 of the above deviations.

  This initial matrix appears in Table 5, after calculating each term.

  f) If you want to quickly obtain an image at the definition of 1 mm, you perform the adjustment calculation, which leads to the result table 6.

  We can resume the process that we used, which consists in creating 18 images from 36 sections made in 36 elementary radiophotographic images and calculating couple by couple the estimated data as indicated previously for the division for the first angle of rotation equal to zero associated with cutting for the second angle of rotation equal to 90.

  The 18 images are then superimposed by rotating each of the images (except the first one) by an angle equal to the opposite of the angle of rotation from the point 0. We show by way of example 4 images, before rotation, obtained for the couples: - 0 - 90 (figure 7), 90 - 180 (figure 8), 180 - 270 (figure 9), 270 - 360 (figure 10) .

  If we allow the tables after rotation, we are led to Tables 11, 12, and 13 (synthesis of the images of Figures 7, 8, 9 and 10).

  To obtain a high definition image, it was necessary to create software having the following functions: - Acquire and record the elementary images and the associated digital data such as those appearing in Figures 3a to 3d.

  graphically defining a slice of the extent of the body to be examined irradiated by the beam of the X-ray source 11 (see FIG. 14).

  - pull in this slice the values associated with the images located in this same plane, for example 36 slices made in 36 elementary radiophotographic images prized at 36 angles of rotation, from 0 to 360 incremented by 10.

  - Perform the calculations to generate, in this case 18 images, by combining the values two by two for rotational angle pairs whose offset is 90.

  - Rotate the images according to the offset of an angle equal to the rotation angle so as to superimpose the 18 images homogeneously.

  - Produce a computer image as shown in Figures 15 and 16 which clearly show the section of a chicken bone examined in the selected cutting plane. The image can be white on a black background or vice versa. A slight halo appears which can be eliminated especially if one takes into account that the axis of rotation and the axis of symmetry of the image are slightly shifted.

  It may be recalled that the term-term calculation results from the formula: H * Cij = Bij + (pj EBij) + m * (ci 1Bij) -n

J

  (E7, Pj ijBIJ) For each pair or pair of angles of rotation offset by 90, we obtained a different image that can be associated by superimposing the different images after rotation according to a method that will be explained here.

  Indeed it is possible thanks to a simple calculation to obtain a simple operator.

  This operator is constituted by the inverse matrix of the following matrix: COSINUS A - SINUS A 0 SINUS A COSINUS A 0 0 0 1 Let us give an example, expressed by the three tables below: Take for example a point of coordinates equal to X respectively = -2; Y = 1; Z = 0.

  The following tables show the calculation process: ANGLE IN DEGREE 270 ANGLE IN RADIAN 4.7124 SINUS -1 COSINUS 0 Initial matrix 0 1 0 -1 0 0 0 0 1 Inverse matrix 0 -1 0 1 0 0 0 0 1 This matrix is multiplied by the vector X, Y, Z, which leads to the following coordinates: X '= - 1; Y' = - 2, Z '= 0 In the case explained above for plates 7, 8, 9, 10, we have been led to note a shift of the axis of rotation and rotates this axis, see Table 17 to find its coordinates in each table shifted by 90, 180, 270 so as to refocus the tables and produce a synthetic image ad hoc.

  Taking into account these shifts, we generated an initial matrix 29 x 29 (Table 11) and a result matrix (Table 12) which results in the multicolored graph of synthesis (Figure 13).

  Table 11 is obtained after rotating the matrices obtained to create and refocus the result tables with rotation angles.

  / 0 for the graph 90 180, / 270 for the graph 90 180, / 180 for the graph 180 270, / 90 for the graph 270 O. These rotations have been simplified by the use of a simple operator that calculates a table after a rotation of 90, taking the transposed matrix of the initial array by multiplying this matrix transposed by the matrix whose all terms are equal to 0, with the exception of the second diagonal, as for example the matrix below .

  0 0 0 1 0 0 1 0 0 1 0 0 1 0 0 0 A rotation of 270 is thus the result of three successive rotations of 90.

  As a result of this experiment, it is not possible to envisage a new type of X or infrared imaging device as described in FIG. 18 and constituted as follows: a turntable 19, on which a person (or an object) to examine is placed. People may be normally upright, but if they are ill, they can be placed on an articulated seat to sit down while being inclined so that the area to be examined can be conveniently accommodated by the seat joints; a beam emitted by an X-ray source 11 with a horizontal axis displaceable along a vertical bracket 23; a detector 13 that can be placed vertically or horizontally along a vertical bracket 25 depending on the size of the region to be examined; a computer 27 which receives, via an analog-to-digital converter, instantaneously the information coming from the detector 13 for each of the flashes of the beam at the moment when the detector 13 and beam are aligned on an axis passing through the axis of rotation 19 of the turntable 31.

  This device can be completed by a second pair of detector beam placed at 90 of the first and operating synchronously; The operation of this system may be as follows: the turntable 31 rotates step by step so that the angle of rotation makes it possible to take images that are just contiguous to one another. If for example the distance from the axis of rotation 19 to the detector 13 is 75 cm, it will perform a complete revolution on a circle of the order of 4.70 m; if the detector 13 is a 23 cm plate, twenty steps will be sufficient in a single turn to obtain elementary images from pairs or pairs taken at 90 from each other; if the pitch is only 6 cm, the circular motion will be cut more precisely in about 80 steps, producing 40 images at the desired definition.

  In the case where one wants to examine small regions vertically one will choose to place the detector 13 in a horizontal position, if one can be satisfied with more extended areas vertically one will place the detector 13 in vertical position, and in some cases we will move the patient to target the area to be analyzed.

  But you can also use other detector configurations for any particular application without the system is actually weighed down.

  The size of the detector 13 can be adjusted by using different detection plates whose cost is reasonable today to obtain a whole range of possibilities.

  Let us recall that a revolution can be carried out without damage in a few seconds and in the examined case one collects 20 or 80 times of the order of 12 million pixels of information making it possible to produce a considerable number of densitometric sections with the definition that the operator can define at will.

  Of course, the higher the definition, the more the calculation time will be.

  The operator will be able to choose his protocol once the images have been stored to obtain at first low or medium definition images and to refine for this or that part of the organism his research while preserving the images created during the scanning, the time to continue his investigation after releasing his patient.

  In any case, for low-resolution images they may appear on the screen a few seconds after the investigation, and if necessary, the investigation may continue for other areas of the body studied.

  The possibility of having two detector beam sets shifted by 90, makes it possible to obtain, in fractions of seconds, images consisting of information taken at the same time, and in about 2 to 3 seconds, the time of a rotation, it is possible to to obtain 10 or 40 images of this or that part of the organism and in this case to have an opportunity to produce a series of images shifted by a few fractions of a second with high scrolling speeds.

  It is thus possible to realize a dynamic view of the interior of an organism by perceiving the movements, to deduce additional information, for example on breathing, or cardiac movements.

  The same device is naturally applicable to the observation of objects to allow: - a non-destructive examination of any object; and even a series of images when one wishes to observe movements inside enclosed objects, in fact a physical object can be rotated rapidly, for example at a speed of 10 or 25 revolutions per second.

  - using a 6 cm wide detector, 40 images per associated revolutions will be created to create a perfect synthesis image and for 25 revolutions per second, obtain a dynamic image equivalent to that produced on a television screen.

  But even for people we can improve the dynamic nature of taking pictures in association several couples working synchronously two by two to obtain a sufficient number of computer images in a turn of one or two seconds of duration.

  Such more complex devices will initially be used by the research.

  The beams X can be replaced by a light beam for example infrared with high penetration power.

  To check the operation of the device described, a routine has been added to the basic software. This routine calculates the standard deviation relative to this point point by point, and by dividing this standard deviation by the reference value, to obtain the level of error relating to the point, then the average of these error levels. therefore also the accuracy of the image as a whole or by particular areas.

  Plate 19 gives the map of the standard deviations.

  The board 20 obtained taking into account this map for the entire central area gives an image whose quality can be evaluated. The results are as follows: the average error for the whole image is 1.07%, the average error for the central zone where the useful part is located is 0.08%, a precision never reached. If you reduce the number of base frames from 18 to 9, the standard deviation is multiplied by 2.

Claims (3)

  1. X-ray or infrared imaging method of a body in which a body to be examined (9) being received by a support (1,3,5,7; 31), the body to be examined is irradiated or illuminated ( 9) with the aid of a source (11) emitting an X-ray beam or a light beam in a direction of propagation, an attenuated intensity is detected in consideration of a crossing of the X-rays or light through the body to be examined ( 9), by means of a beam irradiated or illuminated detector (13), the detected intensities are converted into data for determining an attenuation by the body to be examined by X or light rays, using a digital analog converter (15) is rotated by a rotation angle the support (1-7; 31) mounted movably about an axis of rotation (19) relative to the source (11) and the detector (13) mounted on a frame (21, 23, 25) or rotating the source and the detector mounted on a mobile frame around an ax e of rotation relative to the support, and - with the aid of a computer (27) duly programmed, the following steps are carried out: (1) average the data coming from the conversion of the detected intensities, in n average values at 1 interior of n intervals resulting from a first partition of the data corresponding to a first division of the object to be examined into n elementary layers parallel to the direction of propagation of the beam for a first angle of rotation, and averaging the data coming from the conversion of the detected intensities, in m average values within m intervals resulting from a second partition of the data corresponding to a second division into m elementary layers parallel to the direction of propagation of the beam for a second angle of rotation, preferably different than 90 degrees from the first angle of rotation, the elementary layered cuts forming a grid in nxm zon elementary of a plane of section of the object to be examined (9) defined by the first and the second direction of propagation of the beam for respectively the first and the second angle of rotation, (2) constructing an initial matrix (n, m) with the n and m mean values by assigning to each elementary area a line and column term (Bij) representing an attenuation coefficient defined by the sum of the average value over the interval (i) of the same line as that of the term, divided by the number (m) of columns of the initial matrix and the average value over the interval (j) of the same column as that of the term, divided by the number (n) of rows of the initial matrix (3) adjust the attenuation coefficient in each elementary area by a least squares method taking into account the constraints imposed by the edge values of the average values using the following formula: \ n) (iJBI>) where in this formula, Cij = the value sought Bij = the estimated value initially (n) = the number of rows of the initial matrix (m) = the number of columns of the initial matrix Ein -1Cij = pj for all the values of i, the constraint of the column jm Cij = ci for all the values of j, the constraint of the line i, j = 1 Cij Bij + (i lt * (pj, Bi) + * (ci 77! (i) to obtain a rectified matrix, (4) to repeat steps (1) to (3) for data acquired with different pairs of rotation angles, and (5) averaging the rectified matrices for the different pairs of rotation angles to arrive at a synthesis matrix expressing an image of the attenuation coefficients of the examined body (9) under a definition determined by the grid.   2. Method according to claim 1, characterized in that a large extent of the body to be examined (9) and the detector (13) is irradiated or illuminated in a single control pulse of the source, with the aid of a transmitting tip (29) of the source (11) wide, for example several centimeters in diameter for a cylindrical nozzle, emitting a beam also wide and, - with the aid of the computer (27) duly programmed, the following additional steps are performed (6) recording the data from the conversion of detected intensities across the irradiated or illuminated range of the detector (13), (7) calling the data corresponding to a particular sectional plane by selecting from among the recorded data, those resulting from the conversion of the intensities detected in a slice of the irradiated or illuminated range of the detectors, and (8) performing the steps (1) to (5) from these data from the conversion of s intensities detected in the edge of the irradiated or illuminated area.   A computer program for loading into a computer for carrying out a method according to claim 1 or 2 wherein a test body (9) is received on a support (1,3,5 7, 31), the body to be examined (9) is irradiated or illuminated by means of a source (11) emitting an X-ray or light beam in a direction of propagation, an attenuated intensity is detected in consideration of an X-ray or light beam crossing through the body to be examined (9) by means of a detector (13) irradiated or illuminated by the beam, the detected intensities are converted into data making it possible to determine an attenuation by the body examining X-rays or light with the aid of a digital analog converter (15), and rotating the support (1-7; 31) movably mounted about an axis of rotation ( 19) relative to the source (11) and to the detector (13) mounted on a frame (21,23,25) or to rotate from one to the rotation of the source and the detector mounted on a mobile frame about an axis of rotation relative to the support, the computer program performing the following steps: (1) averaging the data from the conversion of the intensities detected, in n average values within n intervals resulting from a first partition of the data corresponding to a first division of the object to be examined into n elementary layers parallel to the direction of propagation of the beam for a first angle of rotation, and to average the data from the conversion of the detected intensities, in m average values within m intervals resulting from a second partition of the data corresponding to a second division into m elementary layers parallel to the direction of propagation of the beam for a second angle of rotation, preferably not more than 90 degrees from the first angle of rotation, the layered cuttings elementary elements forming a grid in nxm elementary zones of a section plane of the object to be examined (9) defined by the first and the second direction of propagation of the beam for respectively the first and the second angle of rotation, (2) to construct an initial matrix (n, m) with the n and m mean values by assigning to each elementary zone a line and column term (Bij) representing an attenuation coefficient defined by the sum of the mean value over the interval ( i) the same line as the term, divided by the number (m) of columns of the initial matrix and the mean value over the interval (j) of the same column as that of the term, divided by the number (n) of lines of the initial matrix, (3) adjust the attenuation coefficient in each elementary zone by a least squares method taking into account the constraints imposed by the values of borders that are the average values using the form the following: X11 / I * (PJ, Bij) + \ nlr * (ci Bij) - -i nm Cij = Bij + (\ m, * (Ej, pj E JBiJ) where, in this formula, Cij = the value sought Bij = the estimated value initially (n) = the number of rows of the initial matrix (m) = the number of columns of the initial matrix n Ei -1 = pj for all the values of i, the constraint of the column jm Cij = ci for all the values of j, the stress of the line i, = 1 to arrive at a rectified matrix, (4) to repeat the steps (1) to (3) for data acquired with different pairs of angles of rotation and (5) averaging the rectified matrices for the different pairs of rotation angles to arrive at a synthesis matrix expressing an image of the attenuation coefficients of the examined body (9) under a definition determined by the grid.