Radiographic imaging method and apparatus
[DESCRIPTION]
FIELD OF THE INVENTION
The present invention relates to a method and apparatus for
generating an x-ray image of an elongate body in direct radiography
by generating a plurality of partial x-ray images of said elongated
body and by stitching these partial images.
BACKGROUND OF THE INVENTION
In X-ray radiography an x-ray image of an elongate body, such as the
entire spine or the legs of a patient, may have to be obtained.
In Computed Radiography (CR) , such a long length image is generated
by subjecting a number of Imaging Plates (IP), such as photostimulable
phosphor plates, which are organized in a partially
overlapping disposition to an x-ray image of the elongate body. Each
o f the imaging plates carries an image of a part of the elongate
body. After exposure, the individual imaging plates are read out so
as to obtain partial images of the elongate body and finally a long
length image is created by stitching these partial images. Accurate
alignment and measurement can be obtained by superimposing a grid of
radiation attenuating material covering the region to be imaged and
correcting and aligning the partial images to reconstruct the
geometry of said grid. Such methods are described in European
patent applications EP0919856 and EP0866342.
In recent years, Digital Radiography (DR) has become a valuable
alternative for CR. The flat panel detectors (FPD) used in DR are
more costly than the IP's for CR, so an alternative to the one-shot
long length imaging technique of CR using multiple Imaging Plates is
needed. This is achieved by taking plural partial images of an
elongate body by moving the position of the FPD while tilting the Xray
tube or moving the X-ray tube parallel to the FPD.
It is an aspect of the present invention to create an image of the
total elongate body from the partial images in an accurate way
permitting length and angular measurements on the composed image.
SUMMARY OF THE INVENTION
The above-mentioned aspects are realized by a method and apparatus
having the specific features set out in the independent claims.
Specific features for preferred embodiments of the invention are set
out in the dependent claims.
Further advantages and embodiments of the present invention will
become apparent from the following description and drawings.
Long length images are mostly taken to perform length and angle
measurements on the subject across an area larger than a single FPD.
It is therefore important to create an image where the alignment of
the partial images of the subject and the calibration is accurate.
In long length imaging, the elongate image is formed by stitching
partial images of the elongate body which are taken at plural
positions by moving the position of the FPD. In order to support the
subject being imaged, a barrier may be placed between the subject
and the FPD. This barrier has the purpose to stabilize the subject
to minimize movement, to protect the subject from contact with the
moving FPD. It can also be used to attach an object with known
geometry which in accordance with the present invention is used to
align the partial images. Attached to this barrier, multiple rulers
can be applied to determine the region to be imaged and the distance
of the subject to the barrier.
According to the method of the present invention the radiation image
of an object of known geometry is detected and the information on
the geometry of this object in the detected radiation image is used
to geometrically correct the individual partial images before
stitching. The stitch method may again use the geometry of the
detected image of the object of known geometry in each of the
partial image to stitch the images to form one large image.
The method of the present invention thus comprises the steps of
- generating partial radiation images by multiple shot irradiation
and read out of a direct radiography detector, each of the partial
radiation images comprising part of the radiation image of the
elongate body and part of the radiation image of an object of known
geometry superimposed on the radiation image of the elongate body,
- calculating parameters of a geometric transformation expressing
the relation between detected positions of locations of the object
in a partial radiation image and expected positions of the locations
in a partial radiation image,
- projecting the partial images onto a reference plane by applying
the geometric transform to its pixels so as to obtain warped partial
images ,
- stitching the warped partial images so that the image of the
object of known geometric dimensions is reconstructed.
In one embodiment the object of known geometry is a grid
consisting of X-ray attenuating wires which intersect at a given
interval, preferably the interval is 5 by 5 cm.
Alternatively the object of known geometry may consist of a
grid consisting of X-ray attenuating crosses positioned at a given
interval, preferably an interval of 5 by 5 cm.
In a specific embodiment two types of crosses are provided.
All crosses are preferably positioned at a first interval of
preferably 5 x 5 cm and the crosses of a second type are positioned
at a second interval of preferably 10 x 10 cm.
The geometric transformation used in the present invention is
preferably implemented using thin plate spline.
However, alternative implementations are possible, such as
piece-wise linear separable de-skewing.
The method of the present invention is generally implemented in
the form of a computer program product adapted to carry out the
method steps of the present invention when run on a computer. The
computer program product is commonly stored in a computer readable
carrier medium such as a DVD. Alternatively the computer program
product takes the form of an electric signal and can be communicated
to a user through electronic communication.
The invention further discloses a X-ray radiographic apparatus
for image creation of an elongate body which comprises:
- an X-ray imaging unit including an X-ray flat panel detector,
- means for moving said flat panel detector;
- an X-ray generation unit including an X-ray source,
- means for moving said X-ray source,
- an imaging area setting device capable of setting an imaging area
for imaging an elongate body;
- a position determination device for determining a plurality of
positions for the X-ray generation unit and the X-ray imaging unit,
said positions delineating a plurality of partial imaging areas in
said imaging area which overlap with a configured amount,
- an object of known geometry provided between said X-ray imaging
unit and X-ray generation unit, said object comprising parts of Xray
attenuating material distinguishable in images generated by said
X-ray imaging unit;
- at least one control device for controlling said X-ray generation
unit and said X-ray imaging unit so that both units are moved
sequentially to said positions delineating partial image areas and
that partial radiation images of part of said body and part of said
object are generated in each of said positions;
- a processing device for generating an elongate image from the
generated partial images.
A movable diaphragm may be provided to adjust the field of view
of the x-ray source.
The position determining device is capable of generating the
positions and, when applicable, also the diaphragm settings of the
x-ray generation unit and the x-ray imaging unit such that the field
of view area for imaging (the part of the elongate body to be
imaged) is captured by a plurality of images acquired by the flat
panel detector.
The processing unit combines the acquired partial images to
generate a long length image which is an image of the complete field
of view area. For this purpose the processing unit calculates
parameters of a geometric transformation expressing the relation
between detected positions of locations of the object in a partial
radiation image and expected positions of these locations in a
partial radiation image. Next the partial images are projected onto
a reference plane by applying the geometric transform to its pixels
so as to obtain warped partial images. The warped partial images
are then stitched so that the image of said object of known
geometric dimensions is reconstructed
All units can be combined in a single device or implemented in
different devices. In other words, given a desired field of view
area a s input, the computing unit computes the positions and the
diaphragm of the X-ray source and the positions of the flat panel
detector. The controlling unit positions the X-ray source and flat
panel detector and captures the images at each of the positions. The
result is a set of images, each capturing a part of the complete
field of view area and as a union covering the complete field of
view area.
The method and apparatus of the present invention make use of
the radiation image of an object of known geometry present in a
partial image to transform the partial image to a representation
which is similar to an image taken with a flat panel detector
defined in a plane relative to the object of known geometry.
The object of known geometry at least partially covers each of
the field of view areas of each exposure of part of the elongate
body. Based on the known geometry, the processing unit may detect
elements of the object of known geometry and aligns the partial or
transformed images in such a way that the part of the object of
known geometry is correctly represented in the long length image.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 illustrates the parallax effect in the gray area and the
different projection order for the circles if the radiation source
changes position,
Figure 2 illustrates a radiographic image acquisition device wherein
an X-ray generation unit is suspended on the ceiling supporting a
vertical movement of the X-ray generation and the X-ray imaging unit
and supporting a rotation of the X-ray generation unit,
Figure 3 illustrates a radiographic image acquisition apparatus
where the X-ray generation unit is mounted on the floor supporting a
vertical movement of the X-ray generation and the X-ray imaging unit
and supporting a rotation of the X-ray generation unit,
Figure 4 illustrates a radiographic image acquisition apparatus
where the X-ray generation unit and X-ray imaging unit is mounted on
the floor in a single assembly supporting a vertical movement and
rotation of this assembly,
Figure 5 illustrates a radiographic image acquisition apparatus
where the X-ray generation unit and X-ray imaging unit is mounted on
the floor in a single assembly supporting a vertical movement and
rotation of this assembly in combination with a rotation of the Xray
imaging unit,
Figure 6 is a representation of a geometric transformation when the
middle point in a fixed grid is moved somewhat lower and to the
right,
Figure 7 is a representation of the geometric transformation and the
result of a deformed rectangular grid warped back to the original
positions,
Figure 8 is a schematic drawing of a grid consisting of X-ray
attenuating wires which intersect at an interval of 5 by 5 cm which
can serve as an object of known geometry,
Figure 9 is a schematic drawing of a grid consisting of X-ray
attenuating crosses positioned at an interval of 5 by 5 cm wherein
the crosses positioned at an interval of 10 by 10 cm are different
in size than the non overlapping crosses positioned at an interval
of 5 by 5 cm.
DETAILED DESCRIPTION OF THE INVENTION
Acquiring partial images
The method of the present invention is aimed to generate a long
length image suitable for length and angle measurements on the
imaged subject across an area larger than a single flat-paneldetector.
The measurements are preferably accurate in all planes,
not subjective to errors introduced by the parallax effect. Due to
the so-called parallax effect points residing in a different plane
are projected differently if the illumination (radiation) source is
positioned differently. Not only are objects being projected on
different positions in a single reference plane, the projection
order may also change for objects in different planes (see the gray
area with circles on the left drawing in figure 1 ) . To eliminate
this effect, it is advisable to position the illumination source on
the same location when taking different partial images. However,
even when the illumination source position is not fixed, the method
of this invention can still be applied.
To generate an image covering an area bigger than a single flat
panel detector for DR, the following options exist: use more than
one flat panel detector and stack them similar to the method applied
in computed radiography and described e.g. in EP0919856 or use a
special geometry setup for DR such as described in DE102007025448,
or use a single flat panel detector and move it to the different
positions so as to record a multiplicity of partial images together
covering the area of the elongate body. Since the cost of a single
flat panel detector with large dimensions covering the total length
of an elongate body currently is too high, the method of moving the
flat panel detector so as to record a number of partial images of
the elongate body is preferably chosen. Several applicable set-ups
for generating such partial images are shown in figures 2 to 5 .
This choice has two main consequences: multiple exposures are taken
during a certain time interval and the patient must stand away from
the detector to avoid collision. During the time of the exposures,
the patient ideally should not move.
To protect the patient from a collision with the moving imaging
unit, a patient barrier may be placed between the patient and the
imaging unit. If designed properly, the patient barrier described
higher can support the patient to prevent the patient from moving.
When using such a setup, it is clear that all images will be
magnified because of the distance between the patient and the
detector. If this distance is known, this magnification factor is
computed as
ERMF
SOD
where ERMF stands for Estimated Radiographic Magnification
Factor, SID for Source-to-image Distance and SOD for Source-to-
Object Distance where the object represents the patient. The
distance between detector and patient (OID) is given by
OID =SID- SOD .
Often the SOD is not known e.g. because of variations of patient
thickness and variations in the placement of the patient barrier
with respect to the detector.
If an object of known geometry is captured in each of the exposures
generating partial images, the magnification factor can be estimated
from the image content for each of the partial images independently.
Furthermore, if we know how the object of known geometry is
projected on a reference plane close to or in the plane in which the
patient is positioned, we can compensate for all perspective and
other distortions caused by inaccurate alignment or positioning of
the flat panel detector.
Therefore in one embodiment of the present invention the object of
known geometry is in the form of a grid of x-ray attenuating
material which can be used to calibrate the individual images and
transform them to a reference plane, in one embodiment being the
plane of the grid itself.
To minimize the differences between measurements performed in the
grid's reference plane and the measurements of the actual imaged
patient object, this grid should preferably be placed as close as
possible to the patient. To achieve this, the grid is preferably
designed as the object in the patient barrier which supports the
patient. In normal imaging conditions, the patient leans against the
plate containing the grid, as such the distance between patient and
grid is minimal. The design of the grid also allows correct image
stitching as explained below.
Since the partial images are acquired within a certain time
interval, it is preferable to optimize and automate the acquisition
of the partial images. A controlling unit can be used to co-ordinate
the positioning of the X-ray generation unit and the X-ray imaging
unit, the preparation of the X-ray imaging unit, the activation of
the X-ray generation and read-out of the X-ray imaging unit. The
optimization is preferably tuned to minimize the complete time for
the acquisition of all partial images. All other processing related
operations can be postponed to a stage where all images are already
acquired.
A specific embodiment of the image acquisition steps of the method
of the present invention are summarized as follows: first the X-ray
generation unit and X-ray imaging unit are positioned to a default
position which allows the placement of the patient barrier.
Secondly, the patient barrier containing a calibration and stitching
grid is placed to a position close to the detector and the patient
is placed against this patient barrier. Thirdly, after input of the
desired area to be imaged, the partial images are acquired (one
after the other, as fast as possible to prevent patient movement)
and sent to a device which is capable of calibrating and stitching
the partial images to generate an elongate (or complete) image.
Optionally, this device also allows the generated elongate image to
be displayed or corrected before sending it to an archive or
diagnostic workstation.
Transforming the image
This module will transform the partial images read out of the
detector such that they are projected onto a reference plane which
is defined in relation to the object of known geometry. By doing so,
the differences in magnification factors and perspective
deformations between the partial images can be compensated. After
such compensation, the resulting warped images can be stitched
together as if they were recorded with the X-ray source positioned
at the same location for the different partial images. In the
proposed setup where a grid is used, the reference plane is
preferably the plane of the grid itself.
There are many ways to obtain such a transformation, thin plate
splines being one them. It is sufficient to detect reference
locations in the image of the object of known geometry and map these
reference locations to their corresponding location in the abovementioned
reference plane. The resulting thin-plate-spline transform
consists of the affine transformation and coefficients which model
the non-rigid deformation.
Next, to construct the image in the reference plane, the position of
each pixel of this image is mapped to the original image using the
thin-plate-spline and the pixel value at the mapped position in the
original image is extracted. Because this mapped position will not
always correspond with the position of a pixel value, an
interpolation technique can be used to estimate the intensity value.
In figure 6 , a geometric transformation is represented for a grid of
3x3 points where the middle point is moved somewhat lower to the
right. A more realistic configuration is found in figure 7 . Here the
acquired image is represented by the solid gray lines. Under the
assumption that the lines are a representation of a rectangular
grid, the intersections are mapped to their corresponding
coordinates on the grid. The geometric transformation is illustrated
as the dotted lines which maps the gray lines on the solid black
lines in the figure resulting in an almost perfect rectangular
reconstruction of the grid. It is obvious that more specific
deformation models can be used to estimate the deformation of the
grid (e.g. piece-wise linear separable de-skewing as described in
EP0919856) .
If the object of known geometry is a grid consisting of X-ray
attenuating wires which intersect at a given interval, the positions
of the grid lines in the partial images can be extracted by lowlevel
operations such as disclosed in patent application EP0866342.
If a grid consisting of X-ray attenuating crosses is used, a
position x,y in the image with intensity value could be
selected as a possible candidate for a cross if the following
conditions are true
x+ j,y+i * + j,y+i+d
V :0 < ||<
where an be interpreted as the central width of the cross
lines and J 2as the total width, as the length of the lines and
d ,d 2as an indication of the size of a region. It is obvious that all
these parameters can be tuned to increase the robustness of the
detection and that the detection process can be optimized in terms
of memory and computation times with standard optimization
techniques .
Since such a simple detection mechanism may be prune to generate
some false positives, one can accumulate the detected positions by
means of a Hough transform to find the period of the grid and to
reject the false positives. The positions of the crosses can be
further optimized by means of linear regression.
Stitching the images
In the previous section is described how to extract and transform
objects of known geometry into a reference plane.
If the same object is present in all the images, the known geometry
of the object can be used to stitch the images together accurately.
Suppose an element, A , of the object is detected in a first partial
image and an element, B , of the object is detected in a second
partial image. Using the determined deformation fields, both
positions are mapped onto A ' and B ' in the transformed partial
images. Positions A and B ' are now defined in the reference plane.
If the spatial relationship between A ' and B ' in the reference plane
is known, it is easy to position both partial images in such a way
that this spatial relationship is preserved in the combined images.
The object of known geometry can thus be used to combine partial
images or to combine transformed partial images.
It is furthermore possible to combine transformed partial images on
the basis of image information which is not related to the object of
known geometry (e.g. visual combination) . This may be necessary if
the patient has moved between the acquisition of the images.
Method of generating a radiation image of an elongate body
comprising the steps of
- generating partial radiation images by multiple shot
irradiation and read out of a direct radiography detector,
each of said partial radiation images comprising part of the
radiation image of said elongate body and part of the
radiation image of an object of known geometric dimensions
superimposed on the radiation image of said elongate body,
- calculating parameters of a geometric transformation
expressing the relation between detected positions of
locations of said object in a partial radiation image and
expected positions of said locations in a partial radiation
image,
- projecting said partial images onto a reference plane by
applying said geometric transform to its pixels so as to
obtain warped partial images,
- stitching said warped partial images so that the image of
said object of known geometric dimensions is reconstructed.
A method according to claim 1 wherein said object of known
geometric dimensions is a grid consisting of X-ray attenuating
wires which intersect at a given interval.
A method according to claim 2 wherein said interval is 5 by 5
cm.
A method according to claim 1 wherein said object of known
geometry is a grid consisting of X-ray attenuating crosses
positioned at a given interval.
A method according to claim 4 wherein said interval is 5 by 5
cm.
A method according to claim 4 wherein said crosses comprise
crosses of a first and a second type and wherein all crosses
are positioned at a first interval and the crosses of a second
type are positioned at a second interval.
A method according to claim 6 wherein said first interval is 5
by 5 cm and said second interval is 10 by 10 cm.
A method according to claim 1 wherein said geometric
transformation is implemented using thin plate spline.
A method according to claim 1 wherein said geometric
transformation is implemented by piece-wise linear separable
de-skewing .
A radiographic apparatus for generating a radiation image of
an elongate body comprising:
- an X-ray imaging unit including an X-ray flat panel
detector,
- means for moving said flat panel detector;
- an X-ray generation unit including an X-ray source,
- means for moving said X-ray source,
- an imaging area setting device capable of setting an imaging
area for imaging an elongate body;
- a position determination device for determining a plurality
of positions for the X-ray generation unit and the X-ray
imaging unit, said positions delineating a plurality of
partial imaging areas in said imaging area which overlap with
a configured amount,
- an object of known geometry provided between said X-ray
imaging unit and X-ray generation unit, said object comprising
parts of X-ray attenuating material distinguishable in images
generated by said X-ray imaging unit;
- at least one control device for controlling said X-ray
generation unit and said X-ray imaging unit so that both units
are moved sequentially to said positions delineating partial
image areas and that partial radiation images of part of said
body and part of said object are generated in each of said
positions;
- a processing device for generating an elongate image from
the generated partial images .
. radiographic apparatus according to claim 10 comprising a
diaphragm for collimating the output of said X-ray source.
2 .A radiographic apparatus according to claim 10, wherein a
patient barrier unit is positioned ahead of the X-ray imaging
unit supporting said elongate body, the patient barrier unit
being capable of supporting, containing or consisting of the
object of known geometry.
3 . A radiographic apparatus according to claim 10, wherein said
object of known geometry is a grid consisting of X-ray
attenuating wires which intersect at a given interval.
14. A radiographic apparatus according to claim 13, wherein said
interval is 5 by 5 cm.
15. A radiographic apparatus according to claim 10, wherein said
object of know geometry is a grid comprising X-ray attenuating
crosses positioned at a given interval.
16. A radiographic apparatus according to claim 15, wherein said
interval is 5 by 5 cm.
17. A radiographic apparatus according to claim 15 wherein said
crosses comprise crosses of a first and a second type and
wherein all crosses are positioned at a first interval and the
crosses of a second type are positioned at a second interval.
18. radiographic apparatus according to claim 18, wherein said
first interval is 5 by 5 cm and said second interval is 10 by
10 cm.
A radiographic apparatus according to claim 10, wherein said
processing device transforms generated partial images to a
reference plane before combining them to an elongate image.
A radiographic apparatus according to claim 10, wherein said
processing device is arranged to generate said elongate image
from the generated partial images on the basis of the image of
the object of known geometry.
A radiographic apparatus according to claim 10, wherein said
processing device is arranged to
- calculate parameters of a geometric transformation
expressing the relation between detected positions of
locations of said object in a partial radiation image and
expected positions of said locations in a partial radiation
image, and to
- project said partial images onto a reference plane by
applying said geometric transform to its pixels so as to
obtain warped partial images.
A radiographic apparatus according to claim 21 wherein said
transformation is implemented by using a thin plate spline.
A radiographic apparatus according to claim 21 wherein said
transformation is implemented by using a piece-wise linear
separable de-skewing.
A radiographic apparatus according to claim 20, wherein the
processing device combines the transformed partial images
based on image information not related to the object of known
geometry .
A radiographic apparatus according to claim 10 wherein said
processing device combines said partial images based on im a g
information related to the object of known geometry.
26. A radiographic apparatus according to claim 10 wherein said
processing device combines said transformed partial images
based on image information related to the object of known
geometry .
27. A computer program product adapted to carry out the method of
any of claims 1 - 9 when run on a computer.
28. A computer readable medium comprising computer executable
program code adapted to carry out the steps of any of claims 1