Classifications

• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T11/00—2D [Two Dimensional] image generation
  • G06T11/003—Reconstruction from projections, e.g. tomography
  • G06T11/005—Specific pre-processing for tomographic reconstruction, e.g. calibration, source positioning, rebinning, scatter correction, retrospective gating

Definitions

• the invention relates to a method for reconstructing an X-ray image by combining a collection of overlapping vignettes.
• An important, but not exclusive, application of the invention is bone densitometry, that is to say measurement of bone mineral density (DO) of the body, where measurements of the composition are also undertaken.
• DO bone mineral density
• tissue distinguishing lean tissue from fatty tissue.
• the examination can cover large areas of the body.
• BMD is expressed as a mass per unit area which corresponds to the projection, along parallel lines, of the bone mass on a plane related to a unit surface.
• the multiplication of BMD by the bone surface gives the bone mineral content or CMO.
• FIG. 1 shows the normal configuration of the measurements: the radiation comprises a source 1 (point or linear) which is moved with each measurement along the object 2 as well as the network of detectors 3.
• the positions taken are denoted by, lb, le, Id, and 3a, 3b, 3c and 3d.
• the beam 4 of the radiation must include overlapping portions, large enough so that so much of the object 2 is seen at least once, in the positions 4a, 4b, 4c and 4d that are made to take it, and the radiation projection labels, whose positions coincide with those 3a, 3b, 3c and 3d that the detector network 3 successively takes, likewise have portions of recovery. It is therefore impossible to simply juxtapose the thumbnails to obtain the overall image of the object, but on the contrary we must determine the positions of the covering portions on the thumbnails and synthesize the content of these covering portions to reconstruct the picture.
• Figure 2 explains these problems.
• Two vertically spaced details 5a and 5b are found in the covering portion of the labels taken by the detector network 3 at positions 3a and 3b.
• the rays passing through the detail 5a are distant from the gap 6 on the plane of the detector network 3, and those passing through the detail 5b are distant from the gap 7 on the same plane; the differences are different between them, and different from the displacement that had to be imposed on the network of detectors 3 between positions 3a and 3b where the views were taken.
• a good reconstruction of the image with overlapping portions requires combining the measurements associated with each of the details for different thumbnails, which is impossible to do directly since their heights are generally unknown.
• a method of this kind has, however, already been proposed in the art.
• the image is reconstructed by choosing an exact reconstruction of the portions of covering at determined heights, where the important details, and in particular the bones for an X-ray of the body, are likely to be found. To get good results, you must first know the height of these details. Weighting coefficients can favor the results of one or the other of the vignettes according to the position considered on the covering portion. The restitution of the other details of the image is sacrificed.
• Another known method consists in calculating correlations between the overlapping portions of the different labels to evaluate the difference (6, 7 or other) of the rays to be associated in order to synthesize the overlapping portions.
• the correlations depend on preponderant details present on the two overlapping portions and coming from the same place of the object 2.
• the reconstruction of the image is accomplished up to these preponderant details and it is good, if at least these details exist; but as in the previous process, the details located at the other heights will be poorly rendered.
• the invention generally relates to a method of reconstructing a radiographic image of an object traversed by a diverging radiation undergoing a. attenuation, the radiation occupying successive positions having overlapping portions and the attenuation being measured by a network of detectors, on which the radiation is projected and giving vignettes of the image respectively associated with the positions of the radiation and also comprising portions overlap, the method comprising a combination of thumbnails to reconstruct the image, as well as the following steps:
• the attenuation value assigned to each voxel is equal to the sum of the values measured by said associated detector, divided by the number of thumbnails which contribute to giving said associated detector and by the length of each voxel which has been crossed, and the attenuation values of the voxels are combined by a numerical combination on groups of voxels superimposed on the different reconstruction heights.
• the attenuation value assigned to each voxel is obtained by iterative rear projection of the attenuation values measured by the detectors, provisional values being assigned to the voxels and corrected after having been projected onto the detectors calculating differences between sums and the provisional values on projection lines to the values measured by the detectors on said projection lines, and distributing the differences on said projection lines to correct the provisional values.
• FIGS. 4 and 5 are flow diagrams of two modes of the method.
• the object 2 is discretized into elementary volumes or voxels 8 which define reconstruction heights 11.
• the radiation passes through the voxels 8 by rays 9i and 9j, which are several for the voxels 8 belonging to the covering portions, and which originate from respective positions li and lj of the source 1 and project onto respective detectors lOi and lOj which are associated with them for the corresponding positions 3i and 3j of the network of detectors 3.
• the detectors 10 measure attenuation of rays 9 through the whole object 2, and therefore through all the voxels such as 8 which they pass through.
• the voxels 8 project onto a surface which may include several detectors 10 completely, and others partially.
• the system is calibrated to associate with each voxel 8 the detectors 10 on which it projects and distribute over them the proportions of its attenuation.
• these calibration techniques which are quite usual in the art, and will consider voxels 8 projecting completely onto a single detector 10 along a single projection radius for the sake of simplicity of the explanations. .
• the method begins with a step A of general discretization of the object 2 in voxels 8, the layers of which define the reconstruction heights 11 of the image.
• the reconstruction heights 11 will be quite low numerous and the voxels 8 rather parallelepiped, elongated in height, than cubic.
• the following steps B and C consist in placing themselves at a reconstruction height 11 and a determined voxel 8.
• the next step E consists in reading the measurement attenuation of rays 9i and 9j on law detectors and 10j.
• step F an average of these attenuations is made, at least for the voxels 8 belonging to the covering portions and which are therefore crossed by at least two rays 9.
• a weighting can be applied to the different measurements, by granting by example more weight to those which come from substantially vertical rays, which in particular improves the image at the positions of overlap of the vignettes.
• Steps C to E or F are then repeated for all the volumes of the layer considered; after which, in step G, an image of the object 2 is reconstructed.
• This image is an image of the entire object 2, and not just a sectional image at the height considered, since the attenuations measured by the detectors 10 along the rays 9 were supposed to be concentrated at the voxels 8 of the layer at this height.
• step H we return to step B to reconstruct the object 2 at another height, and the cycle of steps C to G begins again with the voxels 8 of the associated layer.
• the images of object 2 have been reconstructed at all heights, they are combined in step H with the hope of obtaining a more exact image.
• Several methods can be envisaged. Perhaps the simplest is to make averages of the images on columns 12 (in FIG. 3) of stacked voxels 8 belonging to different layers, with possibly a weighting to favor the most representative layers. Optionally, you can choose only one of the images that you think is better than the others, or an assembly of several of the images in the places they best represent. All these methods should give better results than those of the prior art which have been described previously.
• the scattered radiation can first of all be subtracted from the measurements before using them.
• Several methods exist for making this subtraction the simplest of which, which is given by way of example, is perhaps to carry out an additional measurement where a screen is interposed between the object 2 and the network of detectors 3 in masking some of the detectors 10.
• the masked detectors 10 are not affected by the direct radiation of the rays 9, but only by the scattered radiation, which is then measured by these detectors and which can be deduced by interpolations for the other detectors.
• the attenuations of a radiation can in general be expressed by a multiplier coefficient of the initial radiation Io less than one and equal to e - ⁇ l, where 1 is the attenuation length and ⁇ the attenuation coefficient characteristic of the material, and which is generally the value we seek to reconstruct the image.
• the detectors 10 directly measure the radiation I which has not been absorbed by the object 2 and which is equal to I 0 e " ⁇ l we can deduce the product ⁇ l, then the value of ⁇ if we divide the values of ⁇ l by the crossing lengths of the object 2 by the spokes 9, after having estimated them by another measurement or having evaluated them geometrically.
• step J a division into blocks is carried out at best in step K.
• block can include the voxels 8 associated with a thumbnail.
• the coefficients mi j of the matrix M represent the contribution of a voxel 8 of index j to the projection along the radius 9 of index i, and can in general be approximated by the length crossed by this radius in this volume.
• the next step L is an evaluation of the attenuation at the voxels 8 of the block considered.
• the first evaluation can be arbitrary, for example at zero values.
• the detector 10 which is associated with it by the ray 9 which passes through it is sought, as in step D of the previous embodiment.
• the next step N is a reading of the measurements of the detectors 10 similar to the step E.
• the determination of the projection radii 9 makes it possible to carry out an evaluation of the projected attenuation values in step O, that is to say - to say that one proceeds to computation MX to evaluate p. By subtracting these evaluated values from the projections from the actual measured values from the same projections, the error made in the evaluation of the values projected in step P is determined.
• the next step Q is a rear projection of this error in the voxels 8 of object 2 in order to correct the evaluated values of the attenuation.
• ⁇ (q) is a relaxation coefficient making it possible not to go too quickly towards a solution which corresponds only to the first blocks and which is between 0 and 2; this coefficient is moreover not uniform in the blocks but may advantageously be higher for the radii substantially vertical, or perpendicular to the detectors 10, in order to give them greater weighting importance, as in the previous embodiment;
• ⁇ bi o c is the transpose of the matrix M for the block considered; the term in the denominator is a standardization term; finally, the terms in parentheses represent the error calculated in step P.
• step R we do the same for the next block, starting again the cycle from step K to step Q, then we return to the first block for a new iteration, until the evaluated attenuations have converged towards a solution, this that is expressed by step R.
• the voxels 8 included in the covering portions of object 2 have been treated in the same way as the others, by simply undergoing more iterations, for each of the blocks to which they belong. .
• the method of the invention makes it possible to reconcile a good quality of restitution of the important details of the object studied with a good overall quality of the image. It is possible to obtain images whose resolution is analogous to the pitch of the detectors 10. We have placed our in the usual situation where the network 3 of detectors accompanies the movement of the radiation 4, but the method could be applied without change with a network of motionless detectors under object 2 and the surface of which would extend to all the projection vignettes.

Landscapes

• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Apparatus For Radiation Diagnosis (AREA)
• Analysing Materials By The Use Of Radiation (AREA)

Applications Claiming Priority (3)

Publications (1)

Family

ID=32406415

Family Applications (1)

Country Status (4)

Citations (1)

Family Cites Families (22)

• 2002
  • 2002-12-23 FR FR0216538A patent/FR2849250B1/fr not_active Expired - Fee Related
• 2003
  • 2003-12-19 EP EP03810006A patent/EP1576544A2/de not_active Withdrawn
  • 2003-12-19 US US10/540,069 patent/US20060251208A1/en not_active Abandoned
  • 2003-12-19 WO PCT/FR2003/050195 patent/WO2004059578A2/fr active Application Filing

Patent Citations (1)

Also Published As

Similar Documents

Legal Events