In computer vision, segmentation refers to the process of partitioning a digital image into multiple regions (sets of pixels). The goal of segmentation is to simplify and/or change the representation of an image into something that is more meaningful and easier to analyze.Image segmentation is typically used to locate objects and boundaries (lines, curves, etc.) in images.
The result of image segmentation is a set of regions that collectively cover the entire image, or a set of contours extracted from the image (see edge detection). Each of the pixels in a region are similar with respect to some characteristic or computed property, such as color, intensity, or texture. Adjacent regions are significantly different with respect to the same characteristic(s).
Thursday, June 19, 2008
3D rendering techniques of CT
Surface rendering
A threshold value of radiodensity is chosen by the operator (e.g. a level that corresponds to bone). A threshold level is set, using edge detection image processing algorithms. From this, a 3-dimensional model can be constructed and displayed on screen. Multiple models can be constructed from various different thresholds, allowing different colors to represent each anatomical component such as bone, muscle, and cartilage. However, the interior structure of each element is not visible in this mode of operation.
Volume rendering
Surface rendering is limited in that it will only display surfaces which meet a threshold density, and will only display the surface that is closest to the imaginary viewer. In volume rendering, transparency and colors are used to allow a better representation of the volume to be shown in a single image - e.g. the bones of the pelvis could be displayed as semi-transparent, so that even at an oblique angle, one part of the image does not conceal another.
Image segmentation
Where different structures have similar radiodensity, it can become impossible to separate them simply by adjusting volume rendering parameters. The solution is called segmentation, a manual or automatic procedure that can remove the unwanted structures from the image.
A threshold value of radiodensity is chosen by the operator (e.g. a level that corresponds to bone). A threshold level is set, using edge detection image processing algorithms. From this, a 3-dimensional model can be constructed and displayed on screen. Multiple models can be constructed from various different thresholds, allowing different colors to represent each anatomical component such as bone, muscle, and cartilage. However, the interior structure of each element is not visible in this mode of operation.
Volume rendering
Surface rendering is limited in that it will only display surfaces which meet a threshold density, and will only display the surface that is closest to the imaginary viewer. In volume rendering, transparency and colors are used to allow a better representation of the volume to be shown in a single image - e.g. the bones of the pelvis could be displayed as semi-transparent, so that even at an oblique angle, one part of the image does not conceal another.
Image segmentation
Where different structures have similar radiodensity, it can become impossible to separate them simply by adjusting volume rendering parameters. The solution is called segmentation, a manual or automatic procedure that can remove the unwanted structures from the image.
Three dimensional (3D) image reconstruction
The principle
Because contemporary CT scanners offer isotropic, or near isotropic, resolution, display of images does not need to be restricted to the conventional axial images. Instead, it is possible for a software program to build a volume by 'stacking' the individual slices one on top of the other. The program may then display the volume in an alternative manner.
Multiplanar reconstruction
Typical screen layout for diagnostic software, showing one 3D and three MPR views
Multiplanar reconstruction (MPR) is the simplest method of reconstruction. A volume is built by stacking the axial slices. The software then cuts slices through the volume in a different plane (usually orthogonal). Optionally, a special projection method, such as maximum-intensity projection (MIP) or minimum-intensity projection (mIP), can be used to build the reconstructed slices.
MPR is frequently used for examining the spine. Axial images through the spine will only show one vertebral body at a time and cannot reliably show the intervertebral discs. By reformatting the volume, it becomes much easier to visualise the position of one vertebral body in relation to the others.
Modern software allows reconstruction in non-orthogonal (oblique) planes so that the optimal plane can be chosen to display an anatomical structure. This may be particularly useful for visualising the structure of the bronchi as these do not lie orthogonal to the direction of the scan.
For vascular imaging, curved-plane reconstruction can be performed. This allows bends in a vessel to be 'straightened' so that the entire length can be visualised on one image, or a short series of images. Once a vessel has been 'straightened' in this way, quantitative measurements of length and cross sectional area can be made, so that surgery or interventional treatment can be planned.
MIP reconstructions enhance areas of high radiodensity, and so are useful for angiographic studies. mIP reconstructions tend to enhance air spaces so are useful for assessing lung structure.
Because contemporary CT scanners offer isotropic, or near isotropic, resolution, display of images does not need to be restricted to the conventional axial images. Instead, it is possible for a software program to build a volume by 'stacking' the individual slices one on top of the other. The program may then display the volume in an alternative manner.
Multiplanar reconstruction
Typical screen layout for diagnostic software, showing one 3D and three MPR views
Multiplanar reconstruction (MPR) is the simplest method of reconstruction. A volume is built by stacking the axial slices. The software then cuts slices through the volume in a different plane (usually orthogonal). Optionally, a special projection method, such as maximum-intensity projection (MIP) or minimum-intensity projection (mIP), can be used to build the reconstructed slices.
MPR is frequently used for examining the spine. Axial images through the spine will only show one vertebral body at a time and cannot reliably show the intervertebral discs. By reformatting the volume, it becomes much easier to visualise the position of one vertebral body in relation to the others.
Modern software allows reconstruction in non-orthogonal (oblique) planes so that the optimal plane can be chosen to display an anatomical structure. This may be particularly useful for visualising the structure of the bronchi as these do not lie orthogonal to the direction of the scan.
For vascular imaging, curved-plane reconstruction can be performed. This allows bends in a vessel to be 'straightened' so that the entire length can be visualised on one image, or a short series of images. Once a vessel has been 'straightened' in this way, quantitative measurements of length and cross sectional area can be made, so that surgery or interventional treatment can be planned.
MIP reconstructions enhance areas of high radiodensity, and so are useful for angiographic studies. mIP reconstructions tend to enhance air spaces so are useful for assessing lung structure.
What are the Artifacts of CT?
Although CT is a relatively accurate test, it is liable to produce artifacts, such as the following.
Example of Beam Hardening
Aliasing Artifact or Streaks
These appear as dark lines which radiate away from sharp corners. It occurs because it is impossible for the scanner to 'sample' or take enough projections of the object, which is usually metallic. It can also occur when an insufficient X-ray tube current is selected, and insufficient penetration of the x-ray occurs. These artifacts are also closely tied to motion during a scan. This type of artifact commonly occurs in head images around the pituitary fossa area.
Partial Volume Effect
This appears as 'blurring' over sharp edges. It is due to the scanner being unable to differentiate between a small amount of high-density material (e.g. bone) and a larger amount of lower density (e.g. cartilage). The processor tries to average out the two densities or structures, and information is lost. This can be partially overcome by scanning using thinner slices.
Ring Artifact
Probably the most common mechanical artifact, the image of one or many 'rings' appears within an image. This is usually due to a detector fault.
Noise Artifact
This appears as graining on the image and is caused by a low signal to noise ratio. This occurs more commonly when a thin slice thickness is used. It can also occur when the kV or mA of the X-ray tube is insufficient to penetrate the anatomy.
Motion Artifact
This is seen as blurring and/or streaking which is caused by movement of the object being imaged.
Windmill
Streaking appearances can occur when the detectors intersect the reconstruction plane. This can be reduced with filters or a reduction in pitch.
Beam Hardening
This can give a 'cupped appearance'. It occurs when there is more attenuation in the center of the object than around the edge. This is easily corrected by filtration and software.
Example of Beam Hardening
Aliasing Artifact or Streaks
These appear as dark lines which radiate away from sharp corners. It occurs because it is impossible for the scanner to 'sample' or take enough projections of the object, which is usually metallic. It can also occur when an insufficient X-ray tube current is selected, and insufficient penetration of the x-ray occurs. These artifacts are also closely tied to motion during a scan. This type of artifact commonly occurs in head images around the pituitary fossa area.
Partial Volume Effect
This appears as 'blurring' over sharp edges. It is due to the scanner being unable to differentiate between a small amount of high-density material (e.g. bone) and a larger amount of lower density (e.g. cartilage). The processor tries to average out the two densities or structures, and information is lost. This can be partially overcome by scanning using thinner slices.
Ring Artifact
Probably the most common mechanical artifact, the image of one or many 'rings' appears within an image. This is usually due to a detector fault.
Noise Artifact
This appears as graining on the image and is caused by a low signal to noise ratio. This occurs more commonly when a thin slice thickness is used. It can also occur when the kV or mA of the X-ray tube is insufficient to penetrate the anatomy.
Motion Artifact
This is seen as blurring and/or streaking which is caused by movement of the object being imaged.
Windmill
Streaking appearances can occur when the detectors intersect the reconstruction plane. This can be reduced with filters or a reduction in pitch.
Beam Hardening
This can give a 'cupped appearance'. It occurs when there is more attenuation in the center of the object than around the edge. This is easily corrected by filtration and software.
Subscribe to:
Posts (Atom)
