Method for reconstructing a tomosynthesis plane of a target organ during tomosynthesis measurements, and associated tomosynthesis apparatus

Abstract

This method for reconstructing a tomosynthesis plane of a target organ (14) is computer-implemented and comprises loading (302) projection images acquired by an X-ray detector (10) of a tomosynthesis apparatus (2), and iterations of steps of calculating (303, 304, 305) images reconstructed from the acquired projection images and a preselected or updated three-dimensional motion of the target organ (14), reprojection images, projection residual images, and updating (306) the three-dimensional motion based on the projection residual images. Each iteration is implemented with the three-dimensional motion updated by the previous iteration. Figure for abstract: Fig 3

Description

Title of the invention: Method for reconstructing a tomosynthesis plane of a target organ during tomosynthesis measurements, and associated tomosynthesis apparatus Technical field

[0001] The technical field of the invention is radiography and, more precisely, tomosynthesis which makes it possible to reconstruct the volume of a target organ, for example the breast of a patient, from a limited number of radiographic projections of the target organ.

[0002] In particular, the present invention relates to a method for reconstructing a tomosynthesis plane of the target organ and a tomosynthesis apparatus. Previous techniques

[0003] Breast tomosynthesis is a three-dimensional imaging technology consisting of acquiring two-dimensional, so-called projection, images of a patient's compressed breast. Each projection image is acquired from a fixed X-ray detector for a particular measurement angle during the movement of an X-ray source.

[0004] The movement of the X-ray source is a restricted angular movement which is accompanied by a limited number of acquisitions, for example less than ten projection images.

[0005] Tomosynthesis involves reconstructing cross-sectional views of the compressed breast in planes parallel to the X-ray detector from the projection images, thereby reducing ambiguity due to tissue superposition in conventional full-field mammography images.

[0006] Patient movement during tomosynthesis, such as breathing, can create artifacts, such as blurred areas, object replications, or vertical smears, in the reconstructed cross-sectional views. Therefore, motion correction is necessary to effectively detect possible radiological features, such as masses or microcalcifications.

[0007] The limited number of acquisitions and the restricted angular movement of the X-ray source make motion correction more difficult.

[0008] Tomosynthesis allows a much smaller amount of data to be collected than a tomography method, the correction of the movement of the target organ in tomosynthesis being therefore more complex than in tomography. The reconstruction operation in tomosynthesis is not a complete problem due to the limited number of acquisitions and therefore of data which characterizes this operation.

[0009] US patent 2021 / 0204899 A1 presents a method for quantifying the movement of the compressed breast during the movement of the X-ray source by comparing an actual location of a detail of interest in projection images with an expected location of the object, but does not give complete satisfaction for the correction of this movement. Statement of the invention

[0010] The present invention therefore aims to overcome all or part of the aforementioned drawbacks and, in particular, to estimate the three-dimensional movement of a target organ during tomosynthesis measurements, to precisely determine this movement, to correct it during the reconstruction of a volume of the target organ, and thus to facilitate the detection and recognition of details of interest in the reconstructed volume of the target organ, and to reduce or at least not increase the exposure to X-rays of the target organ.

[0011] The present invention relates to a method for estimating the movement of a target organ, in particular a breast, during tomosynthesis measurements of the target organ acquired for measurement positions of an X-ray source of a tomosynthesis apparatus provided with an X-ray detector, the method being implemented by computer and comprising the loading of projection images acquired by the X-ray detector, the method further comprising iterations of the steps:

[0012] calculating a plurality of reconstructed images from the acquired projection images and a preselected or updated three-dimensional motion of the target organ, the plurality of reconstructed images forming a reconstructed volume of the target organ;

[0013] calculating a reprojection image for each measurement position of the X-ray source, each reprojection image being calculated from the reconstructed volume and the preselected or updated three-dimensional movement;

[0014] calculating a projection residue image for each measurement position of the X-ray source, each projection residue image being calculated as the difference between the acquired projection image considered and the corresponding reprojection image; and

[0015] updating the three-dimensional movement based on the projection residue images;

[0016] Each iteration of said calculation steps is implemented with the three-dimensional movement updated by the previous iteration.

[0017] Each iteration performed makes it possible to increase the accuracy of estimation of the three-dimensional movement. At each iteration, a reconstructed volume corrected for the three-dimensional movement of the target organ is used to more precisely estimate this three-dimensional movement.

[0018] Advantageously, the iterations are carried out until a convergence criterion of the three-dimensional movement updating step is reached.

[0019] Each reprojection image is calculated as a projection image of the reconstructed volume from a measurement position.

[0020] Each projection residual image is calculated as the difference between the acquired projection and reprojection images that correspond to the same measurement position.

[0021] The preselected three-dimensional movement is freely selected at the initialization of the method.

[0022] According to a first embodiment, the preselected three-dimensional movement is zero at the initialization of the method.

[0023] The first embodiment corresponds to an implementation of the method without a priori knowledge of the movement of the target organ.

[0024] According to a second embodiment, the preselected three-dimensional movement is estimated at the initialization of the method by a visible camera or by a motion sensor.

[0025] The second embodiment allows for faster implementation of the method by having a priori knowledge of the movement of the target organ.

[0026] Advantageously, the preselected three-dimensional movement is decomposed on a preselected kinematic basis defining the movements of the target organ authorized for the calculation steps.

[0027] The preselected kinematic basis is freely selected at the initialization of the method. The kinematic basis defines a solution space for the estimation of the three-dimensional movement performed by the target organ during the tomosynthesis measurements. Alternatively, the kinematic basis could be updated during the iterations of the method.

[0028] Preferably, the kinematic basis is a kinematic basis representative of the possible movements of the target organ during the tomosynthesis measurements.

[0029] Optionally, the method comprises the selection by an operator of a kinematic base adapted to the tomosynthesis measurements of the target organ carried out.

[0030] According to an exemplary implementation, the step of updating the three-dimensional movement comprises, for each projection residue image, the calculation of a displacement field of the target organ making it possible to reduce the norm of the projection residue image considered, in particular by applying a minimization procedure, in particular Gauss-Newton.

[0031] Advantageously, the loading of the acquired projection images comprises a sub-step of selecting a restricted area of ​​the target organ, each iteration said calculation steps being implemented such that each reconstructed image, each reprojection image and each projection residue image corresponds to the restricted area of ​​the target organ.

[0032] By applying the method to the restricted area, the calculation time is reduced, the kinematics of the target organ is simplified and its three-dimensional movement is locally estimated.

[0033] Advantageously, the restricted area is selected so as to include a detail of interest.

[0034] Said restricted zone can be chosen as comprising a detail of interest either for its clinical radiological interpretation, or because it shows an acquisition or reconstruction artifact, or because it indicates the presence of a movement that is detrimental to the interpretation.

[0035] Optionally, the restricted area is selected in at least one acquired projection image, the restricted area being selected so as to comprise a detail of interest.

[0036] Optionally, the restricted area is selected in at least one reconstructed image, in particular calculated from the acquired projection images and a preselected three-dimensional movement, in particular zero, the restricted area being selected so as to include a detail of interest.

[0037] Optionally, the restricted area is selected in at least one projection residue image, in particular calculated from the acquired projection images and a preselected three-dimensional movement, in particular zero, the restricted area being selected so as to comprise a detail of interest.

[0038] Advantageously, the detail of interest is a radiological characteristic of the target organ.

[0039] Optionally, the restricted area can be selected by an operator using a graphical interface.

[0040] The operator may perform the selection of the restricted area based on a visual observation of said acquired projection image, said reconstructed image or said projection residue image. Alternatively, the selection of the restricted area may be performed automatically by an artificial learning method, in particular deep learning, applied to said acquired projection image, said reconstructed image or said projection residue image. The method of selecting the restricted area in said image is susceptible to various variants and modifications which will appear to those skilled in the art.

[0041] Optionally, the method is implemented for a plurality of restricted areas of the target organ and further comprises a step of assembling restricted reconstructed images to form extended reconstructed images of the target organ.

[0042] The extended reconstructed images of the target organ may represent the entire volume of the target organ acquired during the tomosynthesis measurements, or a portion thereof.

[0043] Advantageously, the loading of the acquired projection images comprises a sub-step of modifying the resolution of the acquired projection images used to implement the method, the method comprising repetitions of the loading and of said iterations such that each repetition is carried out with a resolution of each reconstructed image, of each reprojection image and of each projection residue image strictly greater than the resolution used for the previous repetition, the preselected three-dimensional movement used for each repetition being equal to the three-dimensional movement updated by the previous repetition.

[0044] By performing the repetitions for several resolutions, the estimation accuracy of the kinematics of the target organ is increased. By starting with a repetition performed for a coarse resolution, the value of the three-dimensional movement is quickly and roughly updated. By then performing repetitions for increasingly finer resolutions, the three-dimensional movement is incrementally updated to increase the accuracy of its estimation.

[0045] In other words, the method may comprise repetitions of the loading and of said iterations, said repetitions being carried out by loading acquired projection images whose resolution strictly increases from one repetition to the next, the three-dimensional movement updated during a considered repetition being used to initialize the next repetition, the repetitions being carried out up to the measurement resolution.

[0046] Preferably, the method further comprises a step of displaying at least one reconstructed image, in particular at the measurement resolution of the X-ray detector, calculated from the acquired projection images and the three-dimensional movement updated by all the iterations of the method. More preferably, the method further comprises a step of displaying the reconstructed volume of the target organ, in particular at the measurement resolution of the X-ray detector, calculated from the acquired projection images and the three-dimensional movement updated at the end of the method.

[0047] Advantageously, the tomosynthesis measurements are carried out so that a movement of the X-ray source has an angular amplitude relative to the target organ of less than 90°, in particular less than 60°.

[0048] The present invention also relates to an apparatus for tomosynthesis of a target organ, in particular a breast, comprising an X-ray detector and an X-ray source movable between measuring positions, the apparatus comprising in in addition to a processor capable of loading projection images acquired by the X-ray detector at the measuring positions and being provided with:

[0049] a module for calculating a plurality of images reconstructed from the acquired projection images and a preselected or updated three-dimensional movement of the target organ, the plurality of reconstructed images forming a reconstructed volume of the target organ;

[0050] a module for calculating a reprojection image for each measurement position of the X-ray source, each reprojection image being calculated from the reconstructed volume and the preselected or updated three-dimensional movement;

[0051] a module for calculating a projection residue image for each measurement position of the X-ray source, each projection residue image being calculated as the difference between the acquired projection image considered and the corresponding reprojection image; and

[0052] of a module for updating the three-dimensional movement as a function of the projection residual images;

[0053] The processor is configured to implement iterations of said modules such that each iteration of said calculation modules is performed with the three-dimensional movement updated by the three-dimensional movement updating module during the previous iteration.

[0054] The three-dimensional motion updating module can be configured to perform the calculation, for each projection residue image, of a displacement field of the target organ making it possible to reduce the norm of the projection residue image considered, in particular by applying a minimization procedure, in particular Gauss-Newton.

[0055] The processor may be capable of performing a selection of a restricted area of ​​the target organ and implementing each iteration of said calculation modules such that each reconstructed image, each reprojection image and each projection residue image corresponds to the restricted area of ​​the target organ.

[0056] The processor may be capable of performing multiple selections of restricted areas of the target organ and performing assembly of restricted reconstructed images to form expanded reconstructed images of the target organ.

[0057] The processor may be capable of performing the modification of the resolution of the acquired projection images used to implement said modules. The processor may be configured to implement repetitions of said modules such that each repetition is performed with a resolution of each reconstructed image, each reprojection image and each projection residue image strictly greater than the resolution used for the previous repetition, the preselected three-dimensional movement used for each repetition being equal to the three-dimensional movement dimensional updated to the previous resolution. In particular, the resolution used can be increased up to the measurement resolution.

[0058] The tomosynthesis apparatus may comprise a screen. The processor may be capable of displaying on the screen at least one reconstructed image, in particular at a measurement resolution of the X-ray detector, calculated from the acquired projection images and the three-dimensional movement updated by all of the iterations.

[0059] The processor may include a module for selecting the preselected three-dimensional movement of the target organ.

[0060] The tomosynthesis apparatus may further incorporate all hardware, modules and software means for implementing the method previously defined. Brief description of the drawings

[0061] Other aims, characteristics and advantages of the invention will appear on reading the following description, given solely by way of non-limiting example and made with reference to the appended drawings in which:

[0062] [Fig.l] is a schematic view of a breast tomosynthesis apparatus according to an exemplary embodiment of the invention;

[0063] [Fig.2] is a schematic view of a breast tomosynthesis measurement according to a exemplary embodiment of the invention;

[0064] [Fig.3], [Fig.4], [Fig.5], [Fig.6] and [Fig.7] schematically illustrate steps of a method for reconstructing a tomosynthesis plane of a target organ according to exemplary embodiments of the invention. Detailed description

[0065] [Fig.l] schematically represents a tomosynthesis apparatus 2 comprising an imaging system 4 and a processing system 6.

[0066] The imaging system 4 is capable of acquiring tomosynthesis measurements and is provided with an X-ray source 8 and an X-ray detector 10.

[0067] The X-ray source 8 is mobile, here on an arc of a circle 12. Alternatively, the X-ray source 8 could be mobile on a segment.

[0068] The X-ray source is capable of emitting a collimated X-ray beam towards a target organ 14 from a set of measuring positions 16, as illustrated in [Fig. 2]. These measuring positions 16 are distributed over the path of the source 12 which extends according to a limited angular amplitude.

[0069] The X-ray beam irradiates the target organ 14, for example the breast of a patient compressed by a compression pad 18.

[0070] The X-ray detector 10 is arranged on the opposite side to the X-ray source 8 with respect to the target organ 14 and receives the radiation having passed through the target organ 14.

[0071] The X-ray detector 10 comprises a grid of pixels, for example a grid of N rows and M columns of pixels, each pixel being capable of measuring the intensity of X-ray radiation received for each measurement position 16 of the X-ray source 8.

[0072] A tomosynthesis measurement makes it possible to acquire, for each measurement position 16, a so-called projection image having a measurement definition equal to the number of pixels of the X-ray detector 10, for example a definition of M*N pixels. The physical size of the pixels of the X-ray detector 10 defines the measurement resolution.

[0073] Preferably, each acquired projection image undergoes preliminary processing known to those skilled in the art comprising, for example, a correction by a white image of the intensity received by the X-ray detector 10.

[0074] The processing system 6 comprises a processor 20 for processing the tomosynthesis measurements and a screen 22 for interacting with an operator.

[0075] The processor 20 comprises a processing unit and a memory. The memory is capable of storing instructions which, when executed by the processing unit, result in the implementation of a method for reconstructing a tomosynthesis plane of the target organ 14, this method making it possible in particular to estimate the three-dimensional movement of the target organ 14.

[0076] This method is implemented by computer from the tomosynthesis measurements and makes it possible to estimate what the three-dimensional movement of the target organ 14 was during these measurements.

[0077] According to a first exemplary embodiment illustrated in [Fig.3], the method comprises steps 300.

[0078] In a step 301, a three-dimensional movement of the target organ 14 is pre-selected. For example, the three-dimensional movement is initialized with a zero value. This zero value may be a default value recorded by the memory of the processor 20. Alternatively, the tomosynthesis apparatus could comprise a visible camera, or a motion sensor, capable of measuring a movement of the target organ. In this case, the three-dimensional movement of the target organ is selected as the movement measured by the visible camera or by the motion sensor.

[0079] This preselected three-dimensional movement will then be updated by the steps of the method for estimating the kinematics of the target organ 14 during the movement of the X-ray source 8.

[0080] During step 301, a kinematic base is also selected on which the preselected or updated three-dimensional movement is decomposed. This kinematic base defines the three-dimensional movements of the target organ authorized during numerical calculation steps of the method. The search for the kinematics of the target organ is therefore carried out in the preselected kinematic base.

[0081] Preferably, this kinematic basis makes it possible to describe the three-dimensional movements of the target organ 14 during tomosynthesis. For example, the operator selects a kinematic basis capable of describing translations in the plane of the X-ray detector 10 and a rotation around the direction normal to the X-ray detector 10.

[0082] The three-dimensional movement and the kinematic basis can be selected via the screen 22 by the operator, in particular from lists stored in the memory of the processor 20. For example, the operator selects the three-dimensional movement based on a visual observation of projection images at the measurement resolution. If the operator observes, in a projection image, a trail in a particular direction, the operator can, for example, select the three-dimensional movement as comprising at least one translational movement in the particular direction.

[0083] The kinematic basis is independent of the tomosynthesis measurement data. Thus, the operator can modify the kinematic basis considered until a better execution of the process is obtained.

[0084] Then, during a step 302, projection images of the target organ 14 acquired by the X-ray detector 10 are loaded.

[0085] In a step 303, reconstructed images of the target organ 14 are then calculated from the acquired projection images and the preselected or updated three-dimensional movement, for example by an algebraic reconstruction method. These reconstructed images constitute a reconstructed digital volume of the target organ 14 that the operator can consult slice by slice along the direction from the X-ray detector 10 to the source 8. The number of slices of the digital volume is conventionally chosen so that the distance between two successive slices corresponds to approximately 1 millimeter of the target organ. Of course, another number of slices can be used.

[0086] A reprojection image is then calculated, during a step 304, for each measurement position 16 of the X-ray source 8, a projection image having been acquired at each measurement position 16. Each reprojection image is calculated from the reconstructed volume and the preselected or updated three-dimensional movement associated with the measurement positions 16 of the X-ray source 8.

[0087] Each reprojection image is calculated by applying a projection operation to the digital volume, the projection operation describing the accumulation of the attenuation of the X-rays 8 during the acquisition of data from the tomosynthesis measurements from the measurement position 16 considered.

[0088] Then, during a step 305, a projection residue image is calculated for each measurement position 16 of the X-ray source 8. Each projection residue image projection is calculated as the difference between the acquired reprojection and projection images which correspond to the same measurement position 16 of the X-ray source 8.

[0089] Finally, during a step 306, a three-dimensional movement of the target organ 14 associated with the measurement positions 16 of the X-ray source 8 is calculated from each projection residue image.

[0090] For example, for each projection residual image, a displacement field of the target organ 14 is calculated to reduce the norm of the considered projection residual image. The displacement field is decomposed on the preselected kinematic basis and then a least squares problem is solved by applying a minimization procedure to the norm of the considered projection residual image, for example a Gauss-Newton minimization procedure.

[0091] During step 306, the three-dimensional movement is also updated with the value calculated during this step.

[0092] Iterations of the calculation steps 303, 304, 305 and 306 are carried out by replacing for each iteration of these calculation steps 303, 304, 305 the three-dimensional movement concerned by the three-dimensional movement updated during the previous iteration of the updating step 306.

[0093] Thus, at each iteration, a reconstructed volume of the target organ 14 is calculated from the acquired projection images and the three-dimensional movement updated during the updating step 306 of the previous iteration, a reprojection image of this reconstructed volume for each measurement position 16, and a projection residue image for each measurement position 16. From these projection residues, the three-dimensional movement of the target organ 14 is updated.

[0094] These iterations are carried out until a convergence criterion is reached based, for example, on the norm of the relative deviation between two movements of the target organ 14 calculated for two successive iterations, in particular when this deviation becomes less than 10%.

[0095] When the convergence criterion is reached, the memory of the processor 20 records, during a step 307, the three-dimensional movement of the target organ 14 calculated during the iterations of the calculation steps 303, 304, 305 and 306.

[0096] According to a second exemplary embodiment illustrated in [Fig.4], the steps 300 illustrated in [Fig.3] are carried out repeatedly by modifying the resolution of the acquired projection images at each iteration. More precisely, the resolution is reduced, then increased at each iteration until the measurement resolution of the detector is reached. Each iteration of the steps 300 is initialized by the three-dimensional movement of the target organ 14 recorded during the previous repetition.

[0097] Thus, repetitions of steps 300 of the method are carried out. The term iteration is used to describe each iteration of the calculation and update steps 303, 304, 305 and 306 performed at a given resolution. The term repetition is used to describe each repetition of the steps 300 of the method performed with acquired projection images of modified resolution, each repetition therefore comprising several iterations.

[0098] During each repetition of the loading step 302, a sub-step of modifying the resolution of the acquired projection images used for the repetition in question is carried out. The sub-step of modifying the resolution of each repetition is carried out so as to strictly increase the resolution of the acquired projection images compared to the previous repetition.

[0099] Each acquired projection image of modified resolution is, for example, obtained by applying a Gaussian filter to the acquired projection image at the measurement resolution which corresponds to the measurement position 16 considered, then by performing a sub-sampling of the filtered image. For example, a Gaussian filter of variance L is applied to the acquired image at the measurement resolution, then a zone of L*L pixels is replaced by a single pixel having the intensity as the average of the intensity of the L*L pixels of the zone considered. During the repetitions of the steps 300 of the method, an approximation of the movement of the target organ 14 is obtained as a Taylor development with a reduced number of orders.

[0100] Thus, during each repetition of the calculation step 303, each reconstructed image is calculated at the resolution modified during the sub-step of modifying the resolution of the repetition considered.

[0101] Similarly, during each repetition of the calculation step 304, each reprojection image is calculated at the resolution modified during the sub-step of modifying the resolution of the repetition considered.

[0102] Finally, during each repetition of the calculation step 305, each projection residual image is calculated as the difference between acquired projection and reprojection images each at the resolution modified during the sub-step of modifying the resolution of the repetition considered.

[0103] The preselected three-dimensional motion used to initialize steps 300 of each repetition is equal to the three-dimensional motion updated by the previous repetition, i.e., the three-dimensional motion recorded during step 307 of the previous repetition.

[0104] In the example shown, repetitions are carried out for resolutions equal to one tenth, one fifth, one half of the measurement resolution, and the measurement resolution.

[0105] Advantageously, the resolution used for each repetition of steps 300 can be selected by the operator via the screen 22. The operator selects in particular a decrease in resolution by at least a factor equal to an estimated movement of the target organ 14 divided by the size of a pixel. For example, the operator estimates a movement of the target organ 14 of 1 millimeter. If each pixel of the X-ray detector measures 100 microns, the operator selects a resolution equal to one-tenth of the measurement resolution so that the movement of 1 millimeter is detectable. Thus, the three-dimensional movement of the target organ 14 will be estimated accurately.

[0106] Optionally, following the repetition carried out at the measurement resolution, a step is carried out for calculating the reconstructed volume at the measurement resolution from the projection images acquired at the measurement resolution and the three-dimensional movement recorded during step 307 which is carried out during the repetition carried out at the measurement resolution. A step is then carried out for displaying this reconstructed volume at the measurement resolution and corrected for the three-dimensional movement updated by the steps of the method, i.e. by the iterations and the repetitions.

[0107] According to a third exemplary embodiment illustrated in Figures 5 to 7, the steps 300 illustrated in [Fig.3] are carried out for a restricted area of ​​the target organ 14 and therefore for part of the data from the tomosynthesis measurements.

[0108] More precisely, during the loading step 302, a sub-step of selecting a restricted area comprising a detail of interest of the target organ 14 is carried out, for example an area of ​​strictly less than M*N pixels. Then each calculation step 303, 304, 305 is carried out for the selected restricted area.

[0109] Thus, during each calculation step 303, reconstructed images of the restricted area are calculated to form a reconstructed volume of the restricted area.

[0110] Similarly, during each calculation step 304, each reprojection image is calculated from the reconstructed volume of the restricted zone.

[0111] Finally, during each calculation step 305, each projection residue image is calculated as the difference between acquired projection images and reprojection images corresponding to the restricted area.

[0112] In the example illustrated in [Fig. 5], the restricted area of ​​the target organ 14 is selected from a reconstructed image of the entire target organ 14, in particular reconstructed for zero three-dimensional movement of the target organ 14. For example, the operator visually locates a radiological characteristic 501 in the target organ 14 and selects via the screen 22 the restricted area on which to perform the steps 300.

[0113] In the example illustrated in [Fig.6], the restricted area of ​​the target organ 14 is selected from a projection residue image of the entire target organ 14. For example, the operator visually locates a projection residue of interest 601, assimilates it to a radiological characteristic in the target organ 14, and selects via the screen 22 the restricted area on which to perform the steps 300.

[0114] In the example illustrated in [Fig.7], the target organ 14 is broken down into a plurality of restricted zones and the steps 300 are implemented for each restricted zone.

[0115] In a step 700, the assembly of restricted reconstructed images corresponding to each restricted zone is then carried out to form extended reconstructed images of the target organ 14.

[0116] In each of the examples illustrated in Figures 5 to 7, by restricting the area on which steps 300 are carried out, the description of the movement of the target organ is simplified.

[0117] In each of the examples illustrated in Figures 5 to 7, steps 300 can be carried out for each area restricted to the measurement resolution or repetitions of steps 300 can be carried out for each restricted area by increasing the resolution at each repetition and this up to the measurement resolution. Thus, the second exemplary embodiment illustrated in [Fig.4] and the third exemplary embodiment illustrated in Figures 5 to 7 are compatible.

Claims

Claims

1. Method for reconstructing a tomosynthesis plane of a target organ (14), in particular a breast, during tomosynthesis measurements of the target organ (14) acquired for measurement positions (16) of an X-ray source (8) of a tomosynthesis apparatus (2) provided with an X-ray detector (10), the method being computer-implemented and comprising loading (302) projection images acquired by the X-ray detector (10), characterized in that the method further comprises iterations of the steps: calculating (303) a plurality of reconstructed images from the acquired projection images and a preselected or updated three-dimensional movement of the target organ (14), the plurality of reconstructed images forming a reconstructed volume of the target organ (14);calculating (304) the reprojection image for each measurement position (16) of the X-ray source (8), each reprojection image being calculated from the reconstructed volume and the preselected or updated three-dimensional movement; calculating (305) a projection residue image for each measurement position (16) of the X-ray source (8), each projection residue image being calculated as the difference between the acquired projection image considered and the corresponding reprojection image; and updating (306) the three-dimensional movement according to the projection residue images; each iteration of said calculating steps (303, 304, 305) being implemented with the three-dimensional movement updated by the previous iteration (306).;

2. The method of claim 1, wherein loading (302) the acquired projection images comprises a sub-step of selecting a restricted area of ​​the target organ (14), each iteration of said calculation steps (303, 304, 305) being implemented such that each reconstructed image, each reprojection image and each projection residue image corresponds to the restricted area of ​​the target organ (14).

3. The method of claim 2, wherein the restricted area is selected from at least one acquired projection image, the restricted area being selected to include a detail of interest.

4. Method according to claim 2, in which the restricted area is selected in at least one reconstructed image, in particular calculated from the acquired projection images and a preselected three-dimensional movement, in particular zero, the restricted area being selected so as to comprise a detail of interest.

5. Method according to claim 2, in which the restricted area is selected from at least one projection residue image, in particular calculated from the acquired projection images and a preselected three-dimensional movement, in particular zero, the restricted area being selected so as to comprise a detail of interest.

6. A method according to any one of claims 2 to 5, implemented for a plurality of restricted areas of the target organ (14) and further comprising a step of assembling restricted reconstructed images (700) to form expanded reconstructed images of the target organ (14).

7. A method according to any one of claims 1 to 6, wherein loading (302) the acquired projection images comprises a sub-step of modifying the resolution of the acquired projection images used to implement the method, the method comprising repetitions of the loading (302) and of said iterations such that each repetition is performed with a resolution of each reconstructed image, each reprojection image and each projection residue image strictly greater than the resolution used for the previous repetition, the preselected three-dimensional motion used for each repetition being equal to the three-dimensional motion updated by the previous repetition.

8. Method according to any one of claims 1 to 7, further comprising a step of displaying at least one reconstructed image, in particular at a measurement resolution of the X-ray detector (10), calculated from the acquired projection images and the three-dimensional movement updated by all the iterations of the method.

9. Method according to any one of claims 1 to 8, in which the step of updating (306) the three-dimensional movement comprises, for each projection residue image, the calculation of a displacement field of the target organ (14) making it possible to reduce the norm of the projection residue image considered, in particular by applying a minimization procedure, in particular Gauss-Newton.

10. A method according to any one of claims 1 to 9, wherein the preselected three-dimensional motion is decomposed on a preselected kinematic basis defining the target organ motions permitted for the calculation steps.

11. A method according to any one of claims 1 to 10, wherein the preselected three-dimensional motion is zero at the initialization of the method.

12. A method according to any one of claims 1 to 10, wherein the preselected three-dimensional motion is estimated at the initialization of the method by a visible camera or by a motion sensor.

13. A method according to any one of claims 1 to 12, wherein the tomosynthesis measurements are carried out such that a movement of the X-ray source (8) has an angular amplitude relative to the target organ of less than 90°, in particular less than 60°.

14. Apparatus (2) for tomosynthesis of a target organ (14), in particular a breast, comprising an X-ray detector (8) and an X-ray source (8) movable between measuring positions (16), the apparatus (2) further comprising a processor (20) capable of loading projection images acquired by the X-ray detector (8) at the measuring positions (16) and being provided with: a module for calculating a plurality of reconstructed images from the acquired projection images and a preselected or updated three-dimensional movement of the target organ (14), the plurality of reconstructed images forming a reconstructed volume of the target organ (14); a module for calculating a reprojection image for each measuring position (16) of the X-ray source (8), each reprojection image being calculated from the reconstructed volume and the preselected or updated three-dimensional movement;of a module for calculating a projection residue image for each measurement position (16) of the X-ray source (8), each projection residue image being calculated as the difference between the acquired projection image considered and the corresponding reprojection image; and of a module for updating the three-dimensional movement as a function of the projection residue images; the processor being configured to implement iterations of said modules such that each iteration of said calculation modules is carried out with the three-dimensional movement updated by the module for updating the three-dimensional movement during the iteration; previous.

Images