Color Plates Part 3 (Biomedical Image Analysis)

Examples of elevation landscape rendering. The source image (A) is the Shepp-Logan head phantom, with some additive noise in the left ventricle. A scale bar was placed to the right of the phantom. Rendering B was performed using the home-made isometric view that served as this topic's example (Figure 13.6), and rendering C was performed with OpenGL. The red cube in the top left corner of image C is the light source. Elevation landscape rendering is particularly useful for recognizing shapes: From the original image, the intensity function in the scale bar at the right image edge cannot readily be determined. In the landscape renderings, however, the curved nature becomes obvious, and an anti-gamma function can be recognized.


FIGURE 13.7 Examples of elevation landscape rendering. The source image (A) is the Shepp-Logan head phantom, with some additive noise in the left ventricle. A scale bar was placed to the right of the phantom. Rendering B was performed using the home-made isometric view that served as this topic’s example (Figure 13.6), and rendering C was performed with OpenGL. The red cube in the top left corner of image C is the light source. Elevation landscape rendering is particularly useful for recognizing shapes: From the original image, the intensity function in the scale bar at the right image edge cannot readily be determined. In the landscape renderings, however, the curved nature becomes obvious, and an anti-gamma function can be recognized.

Visualization of the brain inside the head. The brain and the head were prepared as separate objects and placed in the scene for rendering. In image A, a section of the head is removed in the head image, whereby the brain is exposed in the removed section. In image B, a clip plane separates the upper section of the skull form the lower, and transparency was used in the upper section.

FIGURE 13.10 Visualization of the brain inside the head. The brain and the head were prepared as separate objects and placed in the scene for rendering. In image A, a section of the head is removed in the head image, whereby the brain is exposed in the removed section. In image B, a clip plane separates the upper section of the skull form the lower, and transparency was used in the upper section.

Three frames from the primate blood vessel animation. Aorta and vena cava are boldly colored (red and blue, respectively), while the spine was assigned a color with less saturation. Although some a priori knowledge was used for segmentation, the course of the blood vessels and the unusual shape of the aorta are clearly visible.

FIGURE 13.12 Three frames from the primate blood vessel animation. Aorta and vena cava are boldly colored (red and blue, respectively), while the spine was assigned a color with less saturation. Although some a priori knowledge was used for segmentation, the course of the blood vessels and the unusual shape of the aorta are clearly visible.

Training of medical students with a virtual patient. The setting is an ophthalmologist's office, and the patient can be asked to perform various tasks, such as telling the number of fingers indicated, or following the hand with his eyes. In this simulation, the hand is controlled by the medical student by means of a three-dimensional positioning device. Simple clicks make it possible to extend a specific number of fingers.

FIGURE 13.13 Training of medical students with a virtual patient. The setting is an ophthalmologist’s office, and the patient can be asked to perform various tasks, such as telling the number of fingers indicated, or following the hand with his eyes. In this simulation, the hand is controlled by the medical student by means of a three-dimensional positioning device. Simple clicks make it possible to extend a specific number of fingers.

Example of an OpenDX visual program. The input image (A) exists in native DX format as created by Crystal Image. In the visual program (B), the String specifies the file name, and the Import operator converts the image into a field. With AutoColor, color is being added to the nodes of the field. Rubbersheet creates the elevation map. Finally, the Image operator renders the object (C). The ColorBar operator is responsible for adding the color scale bar to image C, and the Describe operator describes the data object at the point where it is connected.

FIGURE 14.6 Example of an OpenDX visual program. The input image (A) exists in native DX format as created by Crystal Image. In the visual program (B), the String specifies the file name, and the Import operator converts the image into a field. With AutoColor, color is being added to the nodes of the field. Rubbersheet creates the elevation map. Finally, the Image operator renders the object (C). The ColorBar operator is responsible for adding the color scale bar to image C, and the Describe operator describes the data object at the point where it is connected.

Visualization of a vector field. The visual program (A) is relatively complex because it merges three components: the vector field itself, visualized with the AutoGlyphs operator, and two scalar fields that contain the outline of the flow chamber, visualized as isosurfaces. The lower isosurface (about one-third of the chamber) was made nontransparent, whereas the upper part of the flow chamber was made semitransparent.

FIGURE 14.8 Visualization of a vector field. The visual program (A) is relatively complex because it merges three components: the vector field itself, visualized with the AutoGlyphs operator, and two scalar fields that contain the outline of the flow chamber, visualized as isosurfaces. The lower isosurface (about one-third of the chamber) was made nontransparent, whereas the upper part of the flow chamber was made semitransparent.

1  Show file import info. If active, Crystal Image displays a text window with messages related to the import of TIFF and DICOM files. This option is used to debug import functions and is normally kept inactive.

• Preferred interpolation method. For any operation that needs interpolation, the preferred interpolation method can be selected. The default is bilinear; a better (and slower) choice is bicubic. The spline interpolation methods were developed by Thevenaz et al. (Thevenaz P, Blu T, Unser M. Interpolation revisited. IEEE Trans Med Imaging 2000; 19:739-758) and the implementation should be considered experimental. However, experiments have shown an extremely high quality of iterative interpolation operations.

• Debug message level. If Crystal Image is run from a shell console, debug messages can be printed. A higher number causes more detailed messages to be printed. A good value is 1.

• Undo function. Crystal Image offers two undo stacks, a one-level undo in memory and a 10-level undo on hard disk. For operations on large images, storing images onto the hard disk for undo may cause a noticeable delay.

• Temp path for undo. If the 10-level undo stack on hard disk is selected, a path to a temporary directory needs to be provided. /tmp or /var/tmp is a good

Next post:

Previous post: