Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T12/00—Tomographic reconstruction from projections
  • G06T12/20—Inverse problem, i.e. transformations from projection space into object space
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T2211/00—Image generation
  • G06T2211/40—Computed tomography
  • G06T2211/436—Limited angle

Definitions

• the invention relates to a method of tomosynthesis by lighting an object using an X-ray source.
• the invention finds an application in the field of medical imaging, the field of non-destructive testing of objects and, more generally, in any field implementing the reconstruction of objects scrolling past a source of X-rays.
• Tomosynthesis is a technique which, from a small number of 2D projections (2D for “two-dimensional”) distributed over a limited angular range and acquired on a digital sensor, to reconstruct the 3D volume (3D for “to three dimensions ”) of a studied object.
• FIG. 1 A system for implementing the tomosynthesis method according to the known art is represented, symbolically, in FIG. 1.
• a source 1 of X-rays travels along a linear path 2 in front of an object 3.
• the X-rays 5 which pass through the object 3 are detected by a mobile detector 4 which moves in a direction opposite 6 to the direction of the source 1.
• a computer system (not shown in the figure) collects information from the mobile detector to acquire, process and reconstruct the 3D image of the object.
• the 3D reconstruction is relatively fast but provides relatively fuzzy reconstructions due to the low number of projections and the limited angle character of the 2D projections.
• This patent describes taking into account an object model which makes it possible to strengthen the blood vessels in their transverse direction and to smooth them in their longitudinal direction.
• the processing carried out is not based on any assumption concerning the acquisition geometry or the type of sampling of the reconstruction volume but on assumptions specific to the object.
• the article "Fan-Beam reconstruction from a straight Une of source points" (B. D. Smith, IEEE TMI, vol. 12, num. L, 1993) discloses, to reconstruct the 3D volume, to exploit the linear trajectory of the source in a particular geometry consisting of a series of planes P organized in a fan as shown in FIGS. 2A and 2B.
• a detector plane 4 comprises several parallel lines 7 of detector pixels.
• each plane P passes through the path 2 of the source and a line 7 of detector pixels.
• Each plane P is independent of the other planes.
• Such geometry is only achievable in the. case of a linear trajectory of the source.
• the reconstruction is of an analytical type. The possibility of reconstructing the 3D volume is studied theoretically with an infinite linear trajectory. An exact reconstruction formula is given under these conditions. By passing to a trajectory of the finished linear source, the quality of the reconstructions turns out to be very insufficient and a treatment subsequent to the reconstruction is then necessary.
• the method according to the invention does not have these drawbacks.
• the invention relates to a method of tomosynthesis by lighting an object using an X-ray source having a linear trajectory, the method comprising a step of decomposing the volume of the object into n independent planes forming a fan.
• the method includes a step of anisotropic regularization on each plane.
• the tomosynthesis process comprises an additional step of regularization between planes.
• the tomosynthesis method according to the invention has the advantage of being adapted to the linear path of the source. It is an algebraic processing method particularly well suited to 3D reconstruction with little data. Brief description of the figures
• FIG. 1 represents a block diagram of a system for implementing the tomosynthesis process
• FIGS. 2A and 2B represent 2D planes organized in a fan for regularization in which the X-ray source moves along a linear trajectory
• FIG. 3 represents a tomosynthesis process algorithm according to the invention.
• the method according to the invention comprises, successively, a step E1 of data acquisition, a step E2 of decomposing the reconstruction volume into independent planes and a step E3 of reconstruction and regularization.
• Step E1 is a step known in itself during which an X-ray source and a plane detector move linearly, parallel to each other, in opposite directions, as described above (cf. FIG. 1 ).
• the planar detector consists of a set of lines of detector pixels. With reference to FIG. 1, the space being referenced by the direct trihedron (x, y, z), the trajectory of the source takes place in the direction x and the plane detector is parallel to the plane (x, y).
• Step E2 performs, from the data acquired during step E1, a decomposition of the volume of the object into n planes P forming a fan.
• the n planes P are independent of each other.
• a plane P passes through the path of the source and through a line of pixels of the detector (cf. FIG. 2A).
• the method according to the invention implements step E3 of reconstruction of the 3D volume.
• the reconstruction step is associated with an anisotropic regularization step adapted to the geometry of data acquisition, that is to say to the linear trajectory of the source.
• the anisotropic regularization according to the invention advantageously makes it possible to smooth out certain structures and to enhance others.
• the term "enhance" should be understood as the action of emphasizing or supporting the contrast of a structure.
• noise Three types of artifacts are present during reconstruction: noise, limited angle artifacts and source camera shake. These phenomena are distributed in the three directions of space as follows:
• the anisotropic regularization algorithm according to the invention advantageously makes it possible to treat each of the directions x, y, z differently.
• the anisotropic regularization on a plane P is carried out independently of the regularization on the other planes.
• This anisotropic regularization consists of a smoothing type processing in the x direction and an enhancement type processing in the z direction.
• the regularization on each plane P can be followed by a smoothing type regularization between planes in the direction y.
• the regularization according to the invention is thus adapted to the anisotropy of the reconstruction artifacts linked to the particular acquisition geometry that is the linear and finite trajectory of the source. Indeed, it is according to the directions z and x that the reconstructed volume presents the least good resolution.
• the anisotropic regularization implemented in these directions advantageously makes it possible to very significantly improve the quality of the reconstructed images.
• it is according to the direction y that the acquired data present the best resolution. It follows that a smoothing is sufficient, in the direction y, for improve the quality of reconstructed images.
• the algebraic regularization algorithm implements the minimization of a function J (f), where the variable f represents the object to be reconstructed.
• - p is a projection, i.e. the natural logarithm of the ratio between the information acquired by the detector in the absence of the object and the information acquired by the detector in the presence of the object
• - ⁇ x , ⁇ y , ⁇ z are functions, commonly called potential functions, which determine, according to the respective directions x, y, z, to what extent it is decided or not that an outline of the object to be reconstructed has been detected
• - ⁇ x , ⁇ y, ⁇ z are weighting factors which estimate, according to the respective directions x, y, z, the difference between the projections and data evaluated a priori (data relating to the object to be reconstructed considered as formed of homogeneous zones separated by clear edges)
• ⁇ x , ⁇ y, ⁇ z are minimum grayscale heights of the image in the respective directions x, y, z, from which a contour is accepted ( ⁇ x, ⁇ y and ⁇ z are therefore chosen greater
• 2 is a reconstruction term calculated from the raw acquired data.
• the other three terms of the equation are calculated reconstruction terms which can be treated independently of each other, by modifying the quantities ⁇ x , ⁇ y , ⁇ z , ⁇ ; ⁇ y, ⁇ z , ⁇ x , ⁇ y , ⁇ z -
• the acquisition geometry that is to say the finite linear trajectory of the source, which determines the type of regularization and not the object as is the case according to
• the parameter ⁇ z is chosen greater than ⁇ x and ⁇ y, so as to regularize more strongly in the z direction where the artefacts are strong, because of the linear trajectory of the source.
• the parameter ⁇ z is chosen to be less than ⁇ x and pour y to take more account of discontinuities in the z direction than in the x and y directions.
• the parameters ⁇ y and ⁇ y are chosen to be weak because the resolution along the direction y is - a priori - a good quality resolution.
• the function ⁇ z is preferentially chosen to be concave in order to enhance the reconstructed volume more strongly along the direction z, while the functions ⁇ x and ⁇ y are preferably chosen to be convex.
• two perpendicular linear paths are used for the source.
• the reconstruction then takes place on the basis of a square-based pyramid corresponding to the intersection of two sets of perpendicular fan-shaped planes.
• the method can also comprise a treatment and a correction of the radiation scattered by the object in order to further increase the resolution of the reconstructed image.
• the lines 7 of detector pixels are substantially parallel to the path of the source.
• a resampling of the projections is carried out before the reconstruction step.
• the resampling step can be carried out by interpolation.

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)
• Image Processing (AREA)

Applications Claiming Priority (3)

Publications (1)

Family

ID=27620021

Family Applications (1)

Country Status (4)

Families Citing this family (7)

Family Cites Families (5)

• 2002
  • 2002-02-08 FR FR0201558A patent/FR2835949B1/fr not_active Expired - Fee Related
• 2003
  • 2003-02-06 WO PCT/FR2003/000374 patent/WO2003067525A2/fr not_active Ceased
  • 2003-02-06 EP EP03718844A patent/EP1485873A2/de not_active Withdrawn
  • 2003-02-06 US US10/504,260 patent/US8670601B2/en not_active Expired - Fee Related

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events