ROI Analysis Widget Help


The function of this widget is as a data analysis tool. Pixel values at the crosshair location and pixel values along a line between 2 crosshair locations can be obtained. In addition, histograms statistics on a volume subset of the image data can be obtained. Please see the help for the Edit Crosshairs Widget (from the Edit menu in the DAVE Master Widget) for information about how to position the crosshair.

Data Analysis

The data voxel value at the crosshair location is shown in the green, red and blue boxes near the top of the widget. The green box is for data from the green image and the red box is for data from the red image (if specified on the command line). The blue box for data from the blue image. If you have used the series widget to change which images are red, green, or blue, then the roi widget similarly changes. So whatever is the currently displayed image in the specific color is what gets analyzed. Values in a region of interest can also be examined (see below). By default crosshair values are for integer voxel locations (see "Trilinear Interpolation" below). Also, values are always taken from the highest resolution (i.e., initial) image, regardless of which resolution is currently being rendered in the window. This may cause a discrepancy between what is seen and what is reported in some cases. Also, the integer rounding may mean that when you try to position the crosshair by picking with the cursor (see above) it may not go the the exact cursor location.

Specifying a Region of Interest

Either a line or a box in three dimensions can be specified by selecting two points with the crosshair. The two points either specify the start and end of a line segment or the diagonally opposite corners of a box. Click on "Box" or "Line" to specify which type of region is desired. The box is inclusive of the end coordinates (e.g., a box from 0,0,0 to 2,2,2 has x,y, and z dimensions of 3). Once the crosshair is positioned at the desired location, click on "Accept Pt 1" or "Accept Pt 2" to specify that the current location should be used in the definition of the region of interest. The distance between point 1 and point 2 is displayed in the "Distance" box. This distance is in pixels and assumes the spacing in x,y, and z are all the same (I think). The region of interest can be translated to the current cursor position by clicking on "Trans ROI". If two data sets were given to DAVE, the data set which should be analyzed can be chosen by clicking on "Image 1" or "Image 2".

Inspecting Values along a Line

Once a line has been specified, clicking on "Recalc" will recalculate the voxel values along the line and display them in a graph. Position 0 is at the "Point 1" end of the line segment. The crosshair can be moved to a position along the line by clicking on the arrows under the graph. A yellow triangle points to the position on the graph which corresponds to the crosshair position. The values along the line are calculated as follows. First the line segment is clipped so that any portion of it which lies outside the data volume is thrown away. Then steps of unit pixel length are taken along it. (No z magnification is applied ??). Typically this will result in a position at a noninteger location (i.e., a position in between voxels). This position is rounded to the nearest integer, and the data value at that voxel is then graphed. Since lines are rarely of integer length, there is usually one small small step which is taken at the end to ensure that the endpoint of the line segment is also graphed (so the last data value may be at different spacing along the line than the others).

Trilinear Interpolation

By choosing "ROI Trilinear Interp" from the menu "Region of Interest Histogram" (which pops up when you hold down the right mouse button when the cursor is over that title bar along the top of the widget) you remove the constraint that only the voxel values at integer locations are reported. When interpolation is specified, then the actual noninteger locations are used and data values are obtained by trilinear interpolation from the 8 voxels which surround the point.

Inspecting Values in a Box

Once a box has been specified, clicking on "Recalc" will recalculate a histogram based upon voxel values within the box. As mentioned above, the box is inclusive of the end coordinates (e.g., a box from 0,0,0 to 2,2,2 has x,y, and z dimensions of 3). All statistics are based upon the initial image resolution (so even if you are viewing in lower resolution the numbers will be correct). Num voxels is the number of voxels in the region. ymax at #1 with val #2 means that the histogram has a maximum at intensity #1 (in original image values) with a number of counts of #2. Additional statistics are printed to stdout (so they should show up in whichever window you started DAVE from).


Pressing this will cause the current image to be used to recalculate the image histogram or line intensity values. If the values are current, the recalc button will be green. If the image has changed (via the time series widget) or the position of the endpoints has changed, but the values have NOT yet been recalculated, the button will not be green. Note: the voxel intensity at the cursor position gets updated automatically; the histogram or line intensity values only get updated when recalc is pressed.

NOTE: All voxel values are by default based upon 8 bit scaled values. The 8 bit values are rescaled back to (approximately) the original values by multiplying them by the scale factor and adding back in the black level. This means that the values are only approximate. If the -t command line option is used, then the actual precise data values will be used, not rescaled values.

Copyright 1995 by Lawrence M. Lifshitz and the University of Massachusetts Medical School. All rights reserved.