Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 02566635 2006-11-14
WO 2004/102482 PCT/DK2003/000311
1
Method and system for simulating X-ray images
Field of the invention
The present invention relates to transforming 3D image data to 2D image data
and in
particular to simulating 2D X-ray images from CT data.
Background of the invention
Computed Tomography (CT) scanning of patient anatomy is used by radiologists
for
medical diagnosis. In CT scanning of a patient, a 3D (or volume) image is
generated by
means of an X-ray source which is rapidly rotated around the patient. A large
quantity of
images is obtained resulting in slices of the scanned area, which is
electronically
reassembled to constitute a 3D image of the scanned area. The CT scanning is a
measurement of the amount of X-rays absorbed in the specific volume elements
constituting the 3D image, and each volume element represents the density of
the tissue
comprised in the volume element.
The CT image is a 3D counterpart to the traditional 2D X-ray image. In order
to obtain a
2D X-ray image, an X-ray source is mounted in a fixed position, and a patient
is located
and oriented in-between the X-ray source and a detecting screen. In this
method a
projection of the tissue density along the ray path is acquired.
If a medical diagnosis on the background of CT scanning necessitate surgery, a
surgeon
nearly always requests medical images in the operating theater that display
the particular
medical problem. In some instances, the original CT data of interest is
printed to film, in
the form of cross-sectional slices through the patient anatomy and these films
are provided
to the surgeon. In many other instances, however, the patient is sent to an X-
ray facility
for acquiring 2D X-ray images in addition to the CT images. The request for X-
ray images
is accompanied with a precisely specified patient position for the X-ray
imaging procedure.
Such 2D X-ray projection images are often requested by the surgeons because
the slice
images from CT data are frequently considered as not providing sufficient
information, or
the number of slice images is too large for use in an operation situation.
Sending a patient
to an X-ray facility for obtaining X-ray images is expensive, it takes time
and it exposes
the patient to additional X-ray radiation.
Furthermore, the positioning of a patient in an X-ray apparatus to obtain an
optimal X-ray
image is a skill that technicians must learn. Training is often done on-the-
job, which
sometimes requires re-acquiring X-ray images when the original images do not
show the
CONFIRMATION COPY
CA 02566635 2006-11-14
WO 2004/102482 PCT/DK2003/000311
2
anatomy of interest in the correct way. Obtaining optimal X-ray images may be
difficult
due to organs shadowing the area of interest as well as a proper and precise
positioning of
the X-ray source and the detector plate.
To avoid exposing dental students to X-ray radiation during training of
obtaining X-ray
images of teeth, Umea University has built a virtual training system
(http://www.vrlab.umu.se/forskning/vradiography_eng.shtml). The virtual
training system
comprises a tooth model acquired from a CT scanning. An X-ray image plate may
be
positioned and the virtual model may be oriented, and an X-ray image of the
teeth is
simulated.
Description of the invention
It is an object of the present invention to facilitate access to 2D X-ray
images of a patient
under medical treatment who has undergone CT scanning. Accordingly there is
provided, in
a first aspect, a method of generating a simulated 2D X-ray image from CT data
of a
patient's anatomy, said method comprising the steps of:
- CT scanning the patient to generate corresponding CT data at a first time
instance;
- storing the CT data in a data repository;
- retrieving the CT data at a second time instance after the first time
instance;
- generating a 3D CT image of the anatomy from the CT data; and
- positioning a virtual X-ray source relative to a virtual image plane to
generate the
corresponding 2D simulated X-ray image, whereby the simulated X-ray image may
be
generated from an unrestricted viewpoint.
A patient is CT scanned and the data is stored in a data repository. The
repository of
patient CT data may be accessible to a user through a computer network. For
example, the
CT data may be stored on a computer-based system, such as a client-server
computer
network system. The server may be a central computer, or a central cluster of
computers,
and the CT data may be stored on the server or on a computer to which the
server is
connected. The client may be any type of client, e.g. a thin client, a PC, a
tablet PC, a
workstation, a laptop computer, a mobile hand held device, such as a digital
personal
assistant (PDA) or a mobile phone, or any other type of client. The user may
be a
medically trained person such as a clinician or a surgeon.
The present invention is, however, not limited to implementation on a client-
server type
system. It may be implemented on any type of system, including a workstation
or a PC, or
as a program implemented in connection with the Internet.
CA 02566635 2006-11-14
WO 2004/102482 PCT/DK2003/000311
3
The patient is scanned at a first time instance. The CT data may after this
time instance be
retrieved at a second time instance. The second time instance may be any time
instance
after the first time instance, e.g. it may be during the preparations of an
operation, or it
may be during an operation. While actual X-ray images may be produced before
the
operation, it is impossible to obtain actual X-ray images during an operation.
It is an
advantage of the present invention that a simulated X-ray image may be
obtained at any
time instance after a patient has undergone CT scanning, including a time
instance during
an operation.
The CT image of a patient is acquired from a repository of patient CT data,
for example by
using a software application adapted to visualize CT data, i.e. adapted to
generate a 3D
image from the CT data. The software application is capable of orienting the
3D image with
respect to a viewpoint and a virtual image plane, i.e. rotating the CT image.
The
application may, however, also be capable of other types of manipulation, such
as
zooming, cutting an area, etc. Positioning the virtual X-ray source relative
to a virtual
image plane corresponds to the positioning of the actual X-ray source in an X-
ray facility.
Likewise, the position of the virtual image plane corresponds to the position
of the detector
plane in an X-ray facility, and the virtual image plane may be viewed upon as
a virtual X-
ray detector.
The simulated X-ray image may be generated from a variable and unrestricted
viewpoint.
That is, the virtual X-ray source may be positioned independently with respect
to the CT
object. X-ray images may even be generated from a viewpoint, which is not
possible in a
real X-apparatus. In a real X-ray apparatus a number of viewpoints are not
possible,
because it is not possible to orient the patient in the required way.
A 2D X-ray image may be generated in any given orientation of the 3D CT image
with
respect to the virtual image plane.
The simulated X-ray images can be generated for any patient who has undergone
CT
scanning. After the patient has been scanned, the data may be made available
by storing
the data on a computer system that may be accessed by a user. For example, the
user
may be a surgeon who makes a diagnosis, plans a surgical operation, or a
surgeon who
needs further insight during an operation.
The simulated X-ray image may be simulated using parallel propagating X-rays,
so that
the X-ray image is simulated free of parallax distortion, i.e. an ideal X-ray
image may be
simulated. Alternatively the simulated X-ray image may be simulated using
divergent X-
CA 02566635 2006-11-14
WO 2004/102482 PCT/DK2003/000311
4
rays, so that the X-ray image is simulated with parallax distortion as in
actual X-ray
images.
The CT data of patient anatomy may preferentially be based on data that
conforms to the
Digital Imaging and Communications in Medicine standard (DICOM standard)
implemented
on Picture Archiving and Communications Systems (PACS systems).
The simulation algorithms may be texture based and adapted with special
accumulation,
color, and opacity buffers in a manner that Beer's law is approximated
regarding the
transmission and scattering of photons through physical media.
Additionally, the disclosed method may be used for training of X-ray
technicians via a
computer system that allows the positioning of virtual patient anatomies based
on CT data
sets, and produces virtual X-ray images corresponding to the anatomy position
and other
imaging parameters. Such a system allows for a complete and realistic
training.
According to a second aspect of the invention, a system for generating a
simulated 2D X-
ray image from CT data of a patient's anatomy is provided, the system
comprising:
- a first device and a at least second device, where the first device and at
least second
device are interconnected in a computer network,
- where the first device stores CT data of the patient's anatomy, and
- where the at least second device comprises visualization means as well as
inputting
means capable of accepting request actions,
wherein the patient CT data can be accessed from the at least second device,
so that a 3D
CT image of a patient, and subsequently the simulated 2D X-ray image can be
visualized
on the visualization means comprised in the at least second device.
The first device may be a server and the at least second device a client,
interconnected in
a client-server computer network system. The CT data may be stored on the
first device,
or on a device to which the server is connected by means of a computer network
connection. The at least second device comprises visualization means, such as
a screen on
which data may be visualized both as 3D visualization and 2D visualization.
The at least
second device also comprises inputting means, such as a keyboard and a
computer mouse,
so that request actions, such as keystrokes, mouse movement, mouse clicking,
etc. may
be registered by the at least second device.
CA 02566635 2006-11-14
WO 2004/102482 PCT/DK2003/000311
Brief description of the drawings
Embodiments of the invention will now be described in details with reference
to the
drawings in which:
5 Fig. 1 illustrates the relationship between the 3D CT image and the
simulated X-ray image;
Fig. 2 illustrates the orientation of the 3D CT image relative to a viewpoint
and a virtual
image plane;
Fig. 3 shows a screenshot of a CT object and the corresponding simulated X-ray
image
obtained in connection with a preferred embodiment;
Fig. 4 shows another screenshot; and
Fig. 5 illustrates the difference between simulating an ideal X-ray image
versus simulating
an actual X-ray image.
Detailed description of the invention
The present invention provides a method and system for simulating a 2D X-ray
image from
CT data of a patient's anatomy
In Fig. 1 a block diagram is showing how a CT data may be used either for 3D
reconstruction of the CT data, or for simulating X-ray images. A patient is
first CT scanned
1, after which the data is made accessible, via e.g. a computer network, by
storing the CT
data in a data repository 2. The data repository may be a hard disk or any
other type of
storage medium to which there may be gained access, e.g. via a computer
network. The
CT data may be retrieved 3 from the repository at any time instance after the
data has
been stored in the repository 2. A 3D CT image 4 can be generated from the CT
data by
using a data application such as a program adapted to generate 3D images. A
screenshot
5 showing a 3D CT image is presented. The screenshot 5 shows a foot in a
specific
orientation as well a plate on which the foot was supported during the CT
scanning. A 2D
X-ray image is simulated 6 from the CT data. A screenshot 7 of the simulated X-
ray image
is presented, the support plate is naturally also present in the simulated
image. The
simulated X-ray image is generated in the plane coincident with the image
plane of the
screenshot. The 2D simulated X-ray image is generated as an alternative to an
actual X-
ray image. Obtaining an actual X-ray image requires sending the patient to an
X-ray
facility.
CA 02566635 2006-11-14
WO 2004/102482 PCT/DK2003/000311
6
The 3D CT image may be orientated with respect to a position of the virtual X-
ray source,
i.e. a viewpoint, and a virtual image plane, this is illustrated in Fig. 2.
The 3D CT image 20
may be rotated until a desired orientation is obtained, here exemplified by a
schematic
head viewed in a cross-sectional view 200. The orientation of the CT image 20
is relative
to a virtual X-ray source 24, i.e. the X-ray source used in the simulation of
the 2D X-ray
image. The virtual X-ray source 24 is in a preferred embodiment positioned on
a surface
normal 25 to the virtual image plane 23. But the virtual X-ray source may be
positioned
irrespective of the virtual image plane. By orientating the 3D CT image in
different
positions 21, 22 relative to the virtual X-ray source 24 and the virtual image
plane 23
simulated X-ray images with the patient in these orientations are easily
generated. This is
exemplified here with the head tilted slightly 201, 202 with respect to the
viewpoint and
the virtual image plane. This is especially relevant for obtaining X-ray
images acquired at
slightly different orientations of the patient, a feature that is not easily
feasible with actual
X-ray images. For example, a surgeon may need an X-ray image from a very
specific
angle. But it may be impossible to predict this angle, the position of the
bones may vary
from patient to patient and dislocations of certain bones may be present. It
may therefore
be impossible to predict the optimal angle in order to avoid, or minimize,
shadowing from
various bones. The surgeon may consequently send the patient to an X-ray
facility for
obtaining a series of actual X-ray images in order to evaluate from where an
optimal image
should be obtained. This is followed by that the patient is resent to the X-
ray facility for
obtaining the optimal images. This is a slow and relatively expensive task,
which
furthermore unnecessarily exposes the patient to additional radiation. The
present
invention allows the surgeon, on the background of the CT scanning, to
generate a series
of simulated X-ray images and immediately choose the image that is obtained
from the
optimal viewpoint. This is fast and does not expose the patient to additional
radiation.
The X-ray images are simulated from Computed Tomography by simulation
algorithms that
are texture based and adapted with special accumulation, color, and opacity
buffers in a
manner that Beer's law is approximated regarding the transmission and
scattering of
photons through physical media.
According to Beer's law, the intensity of an X-ray traversing through a
material is
attenuated in the following way:
(I) I = Ij~~t~a~ * EXp (-m*L)
where I,~itiai is the initial intensity of the X-ray, m is the materials
linear attenuation
coefficient, and L is the thickness of the material.
CA 02566635 2006-11-14
WO 2004/102482 PCT/DK2003/000311
7
As the linear attenuation coefficient depends on the material, the intensity
of an X-ray
passing though an object consisting of multiple materials is:
(II) I = Iln~t~a~ * EXp ( f -m(X) dX)
where m(x) is a function that maps a position, x, along the ray to the linear
attenuation
coefficient of the material at the point x of the ray. The function m(x) is
integrated from a
start position at the point source to an end position at the virtual image
plane
The result of a CT scan is a discrete 3D function that maps spatial locations
to Hounsfield
units. Houndsfield units are a standardized and accepted unit for reporting
and displaying
reconstructed X-ray CT values. The system of units represents a line
transformation from
the original linear attenuation coefficient measurements into one where water
is assigned a
value of zero and air is assigned a value of -1 000. If ~w, tta, and w are the
respective
linear attenuation coefficients of water, air and a substance of interest, the
Houndsfield
value of the substance of interest is: H = 1 000 (~, - ~,w)/(~,w - Vita).
Thus, a change of one
Hounsfield unit corresponds to 0.1% of the attenuation coefficient difference
between
water and air, or approximately 0.1% of the attenuation coefficient of water
since the
attenuation coefficient of air is nearly zero. The use of this standardized
scale facilitates
the intercomparison of CT values obtained from different CT scanners and with
different X-
ray beam energy spectra, although the Houndsfield value of materials whose
atomic
composition is very different from that of water will be energy dependent.
The 3D function may now be used to simulate X-ray images of the same object in
the
following way: First the scanned 3D volume is mapped from Hounsfield units
back to linear
attenuation coefficients. A simulated X-ray image may consequently be
generated by
casting rays through the scanned 3D volume, and integrating (ii) along the
path of each
casted ray to calculate the intensity of the X-ray according to Beer's law.
Having estimated the intensity of the X-ray, it is possible to generate an
image by
mapping the X-ray intensities to color/gray scale values.
The 2D X-ray image may be simulated from any viewpoint. Two screenshots are
presented
in Fig. 3 and Fig. 4. In the screenshot 30 in Fig. 3 a 3D reconstruction of a
patient's head
is shown in the left half 31, and in the right half 32 is the corresponding
simulated X-ray
image shown. The X-ray image is simulated with respect to an image plane
coincident with
the image plane of the screenshot. It is possible to generate images from a
viewpoint from
which it is not possible to generate an actual X-ray image since the CT image
may be
CA 02566635 2006-11-14
WO 2004/102482 PCT/DK2003/000311
8
orientated in any orientation (not shown) with respect to the virtual image
plane. Many
viewpoints exist which are not accessible in an actual X-ray apparatus. The
viewpoint may
even be moved within the CT object, so that only the tissue between the
viewpoint and the
virtual image plane contributes to the resulting simulated X-ray image. This
may be
advantageous, e.g. in the case where an image of a feature may not be obtained
because
another feature is shadowing and thereby blocking for the imaging of the
feature of
interest. In a similar manner may a part of the CT image be removed, in this
way the
removed part will not contribute in the simulation of the X-ray image. This is
illustrated in
Fig. 4, which is a screenshot 40 identical to the screenshot 30 in Fig. 3,
except that a
certain part of the CT object has been removed in the left half 41, and that
this part does
not appear in the simulated X-ray image in the right half 42 of the
screenshot.
In Fig. 5 the simulation of an ideal X-ray image versus simulating an actual X-
ray image is
illustrated. A CT object 50 is positioned relative to a virtual X-ray source
51 and a virtual
image plane 52. The X-ray image is simulated either by using divergent X-rays
53, or by
using parallel propagating X-rays 54. The CT object contains a multitude of
features 55,
56, features that will be visible in the simulated image. The features are, in
this example,
identical except that one of the features 55 is positioned further away from
the virtual
image plane 52, than the other feature 56.
In the case that divergent X-rays are used, the simulated X-ray image is
distorted due to
parallax. Objects are broadened and moved according to their position above
the image
plane, the higher above the image plane they are positioned, the more the
feature is
broadened and the further the feature is moved. Thus, feature one 55 is imaged
with a
width b1 57 larger than the width bz 58 of feature two 56, even though the
features are
identical.
In the case that parallel X-rays are used, the simulated X-ray image is not
distorted due to
parallax. In this case feature one 55 is imaged with a width b3 59 equal to
the width b4
60.
Actual X-ray images obtained at an X-ray facility are always parallax
distorted. It may
therefore not always be possible to determine whether a feature is broad due
to the
position of the feature, or due to parallax distortion. One way to resolve
this problem is to
obtain X-ray images from another angle, but this requires resending the
patient to the X-
ray facility. This is expensive, takes time and re-exposes the patient to
radiation. The
present invention provides the possibility to generate an image with and
without parallax
distortion and thereby immediately to determine whether or not the width of an
object of
the X-ray image is largely affected by parallax distortion.
CA 02566635 2006-11-14
WO 2004/102482 PCT/DK2003/000311
9
Although the present invention has been described in connection with preferred
embodiments, it is not intended to be limited to the specific form set forth
herein. Rather,
the scope of the present invention is limited only by the accompanying claims.
