3D position info (in the slice's header) cannot be used because otherwise, as you can very well see, the slices toward the 'End' will get mixed up with those in the 'Middle'!
Sunday, February 18, 2007
Spatial sorting of (DICOM) 2D slices in 3D space.
3D position info (in the slice's header) cannot be used because otherwise, as you can very well see, the slices toward the 'End' will get mixed up with those in the 'Middle'!
Wednesday, February 14, 2007
Jaggies problem.
There are 3 screenshots. In each, the left one is mine, the right one is from OsiriX (a VTK-based application for Mac OS X).
Figure 1. This is the comparison of one of the 23 original 2D slices. Note that when displaying the original slice, there's virtually no quality issue at all!
I build a 3D volume from these 23 slices by repeating each slice's pixels in real-world 3D space till the (voxel) plane thickness becomes equal to the thickness reported in the slice's file header.
Figure 2. This is the coronal transverse section of the volume.
Figure 3. This is the sagittal transverse section of the volume.
My guess is, if there were let's say 1000 such slices (to be displayed in a 512x512 pixel view), the jaggies problem wouldn't manifest... probably because of going from high-resolution data (1000 lines) to low resolution (512 lines).
Figure 1. This is the comparison of one of the 23 original 2D slices. Note that when displaying the original slice, there's virtually no quality issue at all!
I build a 3D volume from these 23 slices by repeating each slice's pixels in real-world 3D space till the (voxel) plane thickness becomes equal to the thickness reported in the slice's file header.
Figure 2. This is the coronal transverse section of the volume.
Figure 3. This is the sagittal transverse section of the volume.
My guess is, if there were let's say 1000 such slices (to be displayed in a 512x512 pixel view), the jaggies problem wouldn't manifest... probably because of going from high-resolution data (1000 lines) to low resolution (512 lines).
Subscribe to:
Posts (Atom)