X-ray or gamma ray systems or devices – Specific application – Computerized tomography
Reexamination Certificate
2003-06-27
2004-11-16
Bruce, David V (Department: 2882)
X-ray or gamma ray systems or devices
Specific application
Computerized tomography
C378S901000
Reexamination Certificate
active
06819735
ABSTRACT:
The present application hereby claims priority under 35 U.S.C. §119 on German patent application number DE 10229113.6 filed Jun. 28, 2002, the entire contents of which are hereby incorporated herein by reference.
BACKGROUND OF THE INVENTION
With modem medical diagnostic methods, such as X-ray computed tomography (CT), image data can be obtained from a measured object that has been examined. As a rule, the measured object that has been examined is a patient.
X-ray computed tomography—designated CT for short below—is a specific X-ray recording method which, in terms of image structure, differs fundamentally from the classical X-ray layer recording method. In the case of CT recordings, transverse slices are obtained, that is to say depictions of body layers which are oriented substantially at right angles to the axis of the body. The tissue-specific physical variable represented in the image is the distribution of the attenuation of X radiation &mgr;(x,y) in the section plane. The CT image is obtained by way of reconstruction of the one-dimensional projections, supplied by the measuring system used, of the two-dimensional distribution of &mgr;(x,y) from numerous different viewing angles.
The projection data is determined from the intensity I of an X-ray after its path through the layer to be depicted and its original intensity I
0
at the X-ray source in accordance with the absorption law:
ln
I
I
0
=
∫
L
μ
(
x
,
y
)
ⅆ
l
(
1
)
The integration path L represents the path of the X-ray considered through the two-dimensional attenuation distribution &mgr;(x,y). An image projection is then composed of the measured values of the linear integrals through the object layer obtained with the X-rays from one viewing direction.
The projections originating from an extremely wide range of directions—characterized by the projection angle &agr;—are obtained by way of a combined X-ray detector system, which rotates about the object in the layer plane. The devices which are most common at present are what are known as “fan ray devices”, in which tubes and an array of detectors (a linear arrangement of detectors) in the layer plane rotates jointly about a centre of rotation which is also the centre of the circular measurement field. The “parallel beam devices”, afflicted by very long measuring times, will not be explained here. However, it should be pointed out that transformation from fan to parallel projections and vice versa is possible, so that the present invention, which is to be explained by using a fan beam device, can also be applied without restriction to parallel beam devices.
In the case of fan beam geometry, a CT recording includes linear integral measured values −1n(I/I
0
) of incoming beams, which are characterized by a two-dimensional combination of the projection angle &agr;&egr;[0,2&pgr;] and the fan angles &bgr;&egr;[−&bgr;
0
,&bgr;
0
](&bgr;
0
is half the fan opening angle) which define the detector positions. Since the measuring system only has a finite number k of detector elements, and a measurement consists of a finite number y of projections, this combination is discrete and can be represented by a matrix:
{tilde over (p)}(&agr;
y
,&bgr;
k
):[0, 2&pgr;)×[−&bgr;
0
, &bgr;
0
] (2)
or
{tilde over (p)}(
y,k
):(1, 2, . . . N
P
)×(1, 2, . . . N
S
) (3)
The matrix {tilde over (p)}(y, k) is called the sinugram for fan beam geometry. The projection number y and the channel number k are of the order of magnitude of 1000.
If the logarithms are formed in accordance with equation (1), then the linear integrals of all the projections
p
(
α
;
β
)
=
ln
I
I
0
=
-
∫
L
μ
(
x
,
y
)
ⅆ
l
(
2
)
are therefore obtained, their entirety also being referred to as the radon transform of the distribution &mgr;(x,y). Such a radon transformation is reversible, and accordingly &mgr;(x,y) can be calculated from p(&agr;,&bgr;) by back-transformation (inverse radon transformation).
In the back-transformation, a convolution algorithm is normally used, in which the linear integrals for each projection are firstly convoluted with a specific function and then back-projected onto the image plane along the original beam directions. This specific function, by which the convolution algorithm is substantially characterized, is referred to as a “convolution core”.
By way of the mathematical configuration of the convolution core, there is the possibility of influencing the image quality specifically during the reconstruction of a CT image from the raw CT data. For example, by way of an appropriate convolution core, high frequencies can be emphasized, in order to increase the local resolution in the image, or by way of a convolution core of an appropriately different nature, high frequencies can be damped in order to reduce the image noise. In summary, therefore, it is possible to state that, during the image reconstruction in computed tomography, by selecting a suitable convolution core, the image characteristic, which is characterized by image sharpness/image contrast and image noise (the two behave in a fashion complementary to each other), can be influenced.
The principle of image reconstruction in CT by calculating the &mgr;-value distribution will not be discussed further. An extensive description of CT image reconstruction is presented, for example, in “Bildgebende Systeme für die medizinische Diagnostik” [Imaging systems for medical diagnostics], 3rd ed, Munich, Publicis MCD Verlag, 1995, author: Morneburg Heinz, ISBN 3-89578-002-2.
However, the task of image reconstruction has not yet been completed with the calculation of the &mgr;-value distribution of the transilluminated layer. The distribution of the attenuation coefficient &mgr; in the medical area of application merely represents an anatomical structure, which still has to be represented in the form of an X-ray image.
Following a proposal by G. N. Hounsfield, it has become generally usual to transform the values of the linear attenuation coefficient &mgr; (which has the dimensional unit cm
−1
) to a dimensionless scale, in which water is given the value 0 and air the value −1000. The calculation formula for this “CT index” is:
CT
index
=
μ
-
μ
water
μ
water
1000
(
4
)
The unit of the CT index is called the “Hounsfield unit” (HU). This scale, referred to as the “Hounsfield scale”, is very well suited to the representation of anatomical tissue, since the unit HU expresses the deviation in parts per thousand from &mgr;
water
and the &mgr; values of most substances inherent in the body differ only slightly from the &mgr; value of water. From the numerical range (from −1000 for air to about 3000), only whole numbers are used to carry the image information.
However, the representation of the entire scale range of about 4000 values would by far exceed the discriminating power of the human eye. In addition, it is often only a small extract from the attenuation value range which is of interest to the observer, for example the differentiation between gray and white brain substance, which differ only by about 10 HU.
For this reason, use is made of what is known as image windowing. In this case, only part of the CT value scale is selected and spread over all the available gray stages. In this way, even small attenuation differences within the selected window become perceptible gray tone differences, while all CT values below the window are represented as black and all CT values above the window are represented as white. The image window can therefore be varied as desired in terms of its central level and also in terms of its width.
Now, in computed tomography, it is of interest in specific recordings to perform organ-specific settings of the image characteristic and, under certain circumstances, organ-specific windowing. For example, in the case of transverse slices through the breast cavity—in which heart, lungs, spinal column are rec
Bruder Herbert
Flohr Thomas
Raupach Rainer
Bruce David V
Siemens Aktiengesellschaft
LandOfFree
Histogram-based image filtering in computed tomography does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Histogram-based image filtering in computed tomography, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Histogram-based image filtering in computed tomography will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3277715