Eppley Institute for Cancer Research 

University of Nebraska Medical Center 


The file explains how HALFSLICE integrates the two dimensional reflection data on each image.
Each reflection is centered in a 21 x 21 pixel box. A default spot size is used (radius 6 pixels) to mask out pixel in the reflection as well as pixels that may intrude in the box from neighboring reflections. An average value of all the remaining pixels is calculated. Pixel more than 10 sigma from this average are removed. A plane is fit to the remaining pixels with the equation:
Background Intensity = A*x*y + B*x + Cy + D
The seedskewing algorithm is used to determine which pixels in the 21x21 box are spots and which are background. The seedskewing algorithm essentially performs a histogram on all the pixels in the box If ther is no spot , bout only background, then the histogram will be symmetric about the mean background value. If there is a spot, then the histogram distribution will be "skewed". Essentially the largest pixels in the box are flagged as spots and removed until the resulting histogram is no longer skewed. The pixels which have been flagged may be considered part of the reflection. The largest contiguous group of pixels near the predicted reflection position is treated as the spot. Constraints are applied so that this spot has a minimum and maximum size. The minimum radius is 2 pixels and the maximum radius is 6 pixels. Finally the resultant group of pixels which have been as a reflection are integrated. From each pixel which is a reflection the analytical background value (as calculated above) is subtracted and the result is summed. This is the final integrated spot intensity.