X-ray or gamma ray systems or devices – Specific application – Stereoscopy
Reexamination Certificate
2000-07-05
2002-04-30
Bruce, David V. (Department: 2882)
X-ray or gamma ray systems or devices
Specific application
Stereoscopy
C378S098000
Reexamination Certificate
active
06381302
ABSTRACT:
FIELD OF THE INVENTION
This invention pertains to the field of three-dimensional imaging and analysis. More specifically, the invention pertains to a system and method for adjusting stereo x-ray images to correct for any vertical or horizontal distortions.
BACKGROUND
For both real-world and computer-generated imaging applications, there is a growing need for display techniques that enable determination of relative spatial locations between objects in an image. Once the spatial relations are established, a user can move through the display space and manipulate objects easily and accurately.
Another related problem is object recognition and localization. One of the basic steps in computer object recognition is to collect as much information as possible about the object by analyzing the image. For example, the space structure of the object can give information that is important for many applications, such as three-dimensional object modeling, vehicle navigation, and geometric inspection. One of the more recent applications is in Computer Aided Diagnosis (CAD), which is a diagnosis made by radiologists utilizing a computer output as a “second opinion”.
A final related problem is binocular stereo, which is the determination of the three-dimensional shape of visible surfaces in a static scene from images taken of the same scene by two cameras or one camera at two different positions. Application of the binocular stereo to X-ray imaging is not easy because there are no visible surfaces on the radiograph and information about different objects can be located at the same areas of the X-ray image. Nevertheless, some of the stereo methods used to analyze standard images could be applied to the analysis of radiograph pairs that were captured using two X-ray sources or by using other stereo imaging techniques.
The approaches used to analyze and resolve the problems described above when examining non-radiograph images can also be applied to radiograph images; provided, however, that the geometry of the stereo digital radiograph system is known. If this is the case, it would be possible to point to an object in the radiograph and calculate an exact location of this object within the radiographed body. It would also be possible to make a disparity map that graphically represents the distance from the X-ray source to every “visible” object in the image. This information could be very important in many medical applications such as surgery, therapy, and related medical applications.
It is straightforward and conventional wisdom that the stereo correspondence problem is a one-dimensional search problem. This is true if the epipolar constraint is known, or selected, from the beginning. In the general case, for example, calibration is used to recover the epipolar geometry accurately. The problem is that even if the imaging geometry is carefully arranged, there are often still errors in the system. This results in corresponding points that are not strictly on the same horizontal lines and distorted vertical positioning. There are other reasons the pixels in one X-ray image may not have matching pixels lying along the same row in the second image and even shifted horizontally. The two major problems result in keystone distortion, vertical parallax and shear distortion.
A well-known effect is keystone distortion (FIG.
2
). Keystone distortion causes vertical parallax in the stereoscopic radiograph due to the baseline of the two X-ray sources being not parallel to the surface of the screen. This is also the case when the stereo radiographs are obtained by the rotation of the object. In one of the radiograph, the image of the square appears larger at one side than at the other. In the other radiograph, this effect is reversed. This results in a vertical difference between homologous points, which is called vertical parallax. The amount of vertical parallax is greatest in the corners of the image.
Another distortion appears when the base line between two X-ray sources and the bottom of the screen are not parallel to the horizon. When the stereo radiographs are obtained by the rotation of the object, this distortion appears when the axis of rotation is not vertical. This distortion also causes vertical parallax as well as shear distortion (FIG.
2
), which influences the correct estimation horizontal location of homologous points.
Numerous algorithms for image matching have been proposed. They can roughly be classified into two categories. In the first category, the algorithms attempt to correlate the gray levels of image patches, assuming that the image patches present some similarity. In the second category, the algorithms first extract salient primitives or feature points from the images, such as edge segments or contours, and then match these features in two views. These methods are relatively fast, because only small subsets of the image pixels are used, but often fail because the chosen primitives cannot be reliably detected in the images.
What is needed is a system and method for quickly determining the geometry of a stereo digital radiograph system enabling quick and accurate correction of distortion in the radiographs such that objects in the radiograph can be located in three dimensional space.
DISCLOSURE OF INVENTION
The present system and method provide a means for quickly and accurately determining the geometry of a stereo x-ray imaging system, enabling the location of objects in the radiograph in three dimensional space. For the purposes of this description, a “body” is the primary item being radiographed. Traditionally, this is a patient at a medical hospital. An “object” is an item that is located within the body such as a bone or joint in the body. A “physical pointer” is an item that absorbs or reflects x-rays such that a mark or point is visibly evident in the radiograph when captured by a digital radiograph system.
The method begins by establishing an initial correspondence using one or more physical pointers around or inside the body. For example, the pointers could be steel balls that make a distinct white spot on the radiographs. Alternatively, other distinctive pointers could be used, for example objects within the body such as bones or joints.
Once the images (
410
) of the body are complete, the physical pointers are used to estimate the epipolar geometry and horizontal distortions of the radiographs. Depending on the distortion estimates, each radiograph is transformed by adjusting it vertically and horizontally. The classical epipolar stereo matching technique can then be used to calculate a disparity map of recognizable objects within the body. Finally, standard geometric properties can be used to calculate the three dimensional location of the objects inside the body.
REFERENCES:
patent: 4669048 (1987-05-01), Ackermann et al.
patent: 6047080 (2000-04-01), Chen et al.
Talukdar, A. et al.; Modeling and Optimization of Rotational C-Arm Steroscopic X-Ray Angiography, IEEE Transactions on Medical Imaging, vol. 18, No. 7, pp. 604-616, (Jul. 1999).
Woods, A. et al.; Image Distortions in Stereoscopic Video Systems, SPIE, vol. 1915 Stereoscopic Displays and Applications IV, pp. 36-48, (1993).
LandOfFree
Computer assisted 2D adjustment of stereo X-ray images does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Computer assisted 2D adjustment of stereo X-ray images, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Computer assisted 2D adjustment of stereo X-ray images will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2868984