sMILX Help

Imaging

This page describes the image processing capabilities of sMILX. A image in sMILX is defined as a set of voxels/pixels arranged as a rectangular volume that has an origin, spacing of each axis and orientation of the volume. An important consideration of the image window is that the original volume is maintained as an ITK image. All processing is applied to the ITK image unless otherwise stated. This is to ensure stability and correctness of the processing as much as possible. Visualisation is done directly with VTK.

There are a few batch operations available in the Image menu, where the images are processed by the order in which they were opened:

Add Images - Sum all (open, unminimised) images in the current tab together to create a single summation image.
Average Images - Sum all (open, unminimised) images in the current tab together and divide by the number of these images to create a single average image.
Difference Images - Subtract all (open, unminimised) images from the first opened image and create a single difference image.
Convolve Images - Convolve all (open, unminimised) images to the first opened image and create a single convolution image.

Note: The above also applies to vector images if ITK 4 was used to build sMILX. Convolution is done in nlogn complexity, if ITK 4 was used to build sMILX.

The remaining options are mostly available through the right click menu and are described below:

Compare - Click on this for every image that your want to show the ckecboard for or to overlay together. (Work in Progress)
Link Windows - Link all the windows so that the camera of the view is always oriented exactly the same in real space in all windows of the tab.
Switch to - Switch to window given in the open windows list.
Import View from - Import the 'actors' (i.e. the display elements, such as the slice or surface or field) to this current display. This can be used for overlaying surfaces on slices etc. Note that the imported actor will be 'in front' of the current actors, useful to know when the actors exactly overlap and one cannot be seen, i.e. the order makes a difference when the objects are in exactly the same space.
Operations - process the voxel/pixel values of the image:

Rescale Inensities - rescale the pixel values to a new provided range.
Smooth (Anisotropic Diffusion) - Smooth the voxels using the Gradient Anistropic Diffusion (parameters in log window).
Smooth (Gaussian Convolution) - Gaussian blur the image.
Smooth (Median) - Median filter the image.
Gradient Magnitude - Show the edges within the image using the Gradient magnitude of the image.
Sobel Edges - Use the three Sobel edge convolution kernels to show the edges in the image.
Normalize - Statistically normalise the image to unit variance and zero mean.
Math Histogram to - match the histogram of the current image to the one provided.

Thresholds - Apply thresholds to the voxel values in various ways:

Otsu Multiple Threshold - Use histgoram based method to label the image in the number of labels + background provided.

Transforms - Transform the image in vairous ways:

Match Info to - match the origin, spacing and orientation of the current image to the one provided.
Resample - place the current image into the image that was provided (as origin, spacing and orientation etc.). Uses B-spline interpolator.
Resample as Labelled Image - same as 'Resample' but with the nearest neighbour interpolator.
Mask Image - mask the image with the mask image provided.
Mask and Crop Image - mask the image with the mask image provided then crop to the bounds of the mask as well.

Display as Volume Rendering - show the image as a volume plot using ray-casting
Display Slice as Surface Plot - show the slice voxel values as a surface plot. The height of the plot is the voxel values.
Display Iso-surface - Create a iso-contour of the image voxels and show it as a polygonal surface (generates a model window).
Image Information - output the origin, spcaing etc. of the image to the log window.
Interpolation - Interpolate the image display
Orientation - Apply the orientation as a transform matrix to the image volume
Contouring - (Work in Progress).