BEAD_FINDER

Overview

BEAD_FINDER is the first step in automatic/interactive bead alignment. This program finds and ranks bead-like features in a data stack, using an adaptive amplitude threshold approach:

  1. Compute the local background image from the local mean or local maximum.
  2. Compute relative height image weighted by a saturation factor. The relative height is wt * (1 - (raw data) / (local background)) where wt is one when the raw data value is greater than or equal to the background mean otherwise wt is wt = exp[-alpha * (mean_sec -local_ref) / mean_sec] where the constant alpha is chosen such that wt equals the saturation factor when the raw data value is the minimum of the background image.
  3. Find the threshold that gives the desired percentage of qualified pixels when a binary image is created by applying an amplitude threshold to the relative height image.
    1. Calculate the histogram of relative height image.
    2. Find the threshold that gives a fraction, f, of the pixels greater than the threshold percentage. How f is controlled is determined by the value of the bArea: mult_pct input parameter.
  4. Calculate the binary thresholded image by assessing each pixel. If the value in the relative height image is greater than the threshold and the pixel has two or more neighbors whose values are above the threshold than the binary image is set to one at that location; otherwise, it is set to zero.
  5. Find bead-like features in the binary image. A bead-like feature satisfies: Then, after a list of such features are found,
    1. If the number of bead-like features found is less than minimum expected number, the two criteria will be loosened to find more. Several such trials are repeated if needed.
    2. If the number of beadk-like features found is greater than the maximum expected number, the two criteria are tightened to find fewer than the maximum. Several such trials are repeated if needed.
    3. Rank each bead-like feature according to its area minus the expected area.
    4. Rank each bead-like feature according to its shape factor.
    5. Score each bead-like feature using these rankings by score = area_rank + shape_rank
    6. Rank the scores and choose up to the maximum expected number of beads with the top scores.

Here's an example command file for BEAD_FINDER:

     (time bead_finder /oahu1/weiping/cent-10k.stk \
     /oahu1/weiping/cent-10k.pl_one -beadrad=3.5 -area_lub=.4:3 \
     -shape_lb=.4 -nbd_minmax=4:50 -barea_pct_mult=3 -size_lbck=-1 \
     -iv=58:58:1 -merge_plfile=none -lbckg_mm=0 -rsat=1 \
     -othrfile=/oahu1/weiping/cent-10k.stk_thr \
     -obinfile=/oahu1/weiping/cent-10k.stk_bin )

Parameters

IdatFile | OplFile | NX:NY:NV | bead_radius | area: lb,ub | shape: lb | nbd_minmax | bArea: mult_pct | lBckg: size | iv | Imerge_plFile | lbckg_mm | rsat | res | ppmin | OthrFile | ObinFile

Related Priism Topics

Priism | Alignment | Bead matcher | Bead chaser | Bead align | Bead align (interactive) | Bead list editor | Reconstruction


IdatFile

This is the name of the file containing the raw projection data stack from the CCD (i.e. measured in terms of electron counts; data stacks with the contrast inverted can not be handled).

Return to the list of parameters


OplFile

OplFile is the point list file that stores the coordinates of found bead-like features on the projections and the corresponding ranking parameters. 7 parameters are saved for each feature:

Return to the list of parameters


NX:NY:NV

The first two values are, respectively, the x and y dimension of the projections. The third value is the number of projections in the data stack.

Return to the list of parameters


bead_radius

This is the expected bead radius in pixels. It sets the expected bead area, area0, which is used to identify bead-like features in binary images. To specify this parameter from the command line, use -beadrad=r; the default value is 3.5 pixels.

Return to the list of parameters


area: lb,ub

Sets the range on the area for a bead-like feature. The first value, arealb in the field is the lower bound on the area; the second value, areaub is the upper bound. Both are expressed as multiples of the expected area, area0, which is set by bead_radius. Thus, a bead-like feature will have an area in square pixels between arealb*PI*bead_radius^2 and areaub*PI*bead_radius^2.

These parameters are set on the command line with -area_lub=arealb:areaub. The default values are 0.4 for the lower bound and 3 for the upper bound.

Return to the list of parameters


shape: lb

The second criteria that bead-like features must meet is that the quantity (the shape factor), 4*PI*(feature area)/(feature circumference)^2, must exceed a threshold. This parameter sets that threshold. By its definition, the shape factor falls in the range [0, 1], and a circle has a shape factor of 1.

On the command line, the threshold for the shape factor is set by -shape_lb=lb. The default value for lb is 0.4.

Return to the list of parameters


nbd_minmax

These two values are, respectively, the minimum and maximum number of beads that are expected on each section. The program will attempt to find more beads if it locates less than the minimum number or reject possible bead-like features if it finds more than the maximum, It does this by dynamically adjusting the two criteria for a bead, the area and shape factor.

The bounds on the number of beads in each section are set from the command line with -nbd_minmax=min:max.

Return to the list of parameters


bArea: mult_pct

This parameter set the upper limit on the area which is non-zero in the binary image. If the limit, u, is greater than one, then the limit imposed on the area is u*(maximum number of beads)*(expected bead area). If u is less than one, then it is used directly as the fraction of each projection which can be non-zero.

The option -barea_pct_mult=u sets the limit from the command line. The default value of u is two.

Return to the list of parameters


lBckg: size

This parameter sets the length (in pixels) of one side of a square region about each pixel which is used to estimate the local background (the difference between the raw projections and the local background gives the relative height image used for locating beads).

The default value, which is the larger of 3 or eight times the bead radius, is used whenever this parameter is less than 3 or is not specified. On the command line, the box size is set by -size_lbck=box_size.

Return to the list of parameters


iv

These three values, which are the index of the first projection to use, the maximum index which can be used, and the index increment, can be used to limit which projections BEAD_FINDER processes. This is frequently useful when experimenting with different thresholds. The default is to process all the projections so the index of the first projection is zero, the maximum index is the number of projections minus one, and the increment is one.

The range of projections to use is set from the command line with -iv=first:max:step.

Return to the list of parameters


Imerge_plFile

The beads found in a run of BEAD_FINDER will be merged with those stored in this file. That is, for the bead list of any section, if BEAD_FINDER is run on that section, the corresponding bead list is saved to the output .pl file; if BEAD_FINDER is not run on that section, the results for that section (if any) are copied from the file specified by merge_plfile.

Merging is normally done when the bead feature criterion are good for most sections, but a few sections need special handling with different parameters.

On the command line, the file with which to merge the results is specified with -merge_plfile=file_name. If none is used as the file name (this is the default), no merging is done.

Return to the list of parameters


lbckg_mm

With the local bckg: mean option, a mean over a local region is used to estimate the background value. This is the default. With the local bckg: max option, the maximum value in a local region is used to estimate the background. The size of the local region is set by lBckg: size.

On the command line, use -lbckg_mm=0 to use the local mean and use -lbckg_mm=1 to use the local maximum.

Return to the list of parameters


rsat

This parameter is the saturation value used to weight the difference of pixel value and the estimate of the background level. The weighted differences at each pixel form the relative height image in which beads are located. In detail, the relative height image at a point is wt * (1 - (raw pixel value) / (local background)). Where the weighting factor, wt, is one when the raw data value is greater than the mean of the projection and is exp[-alpha*((projection mean)-(raw value))/(projection mean)] where alpha is set such that wt is equal to rsat when the raw pixel value is equal to the smallest value in the estimated background image. When rsat is 1 (the default) no weighting is applied.

The saturation level is set on the command line with -rsat=level.

Return to the list of parameters


res

This parameter sets which resolution to use from the input tilt series if the input tilt series has multiple resolutions available. The resolutions are numbered from zero with zero as the highest resolution and increasing values correspond to decreasing resolution. The resolution to use is set on the command line with -res=ires.

Return to the list of parameters


ppmin

This parameter sets a lower bound on intensity values included in the calculation of the background statistics. The default value is 100 to match what previous versions implicitly used. The command line parameter to set the lower bound is -ppmin=bound.

Return to the list of parameters


OthrFile

This parameter sets the name of the image file which will be filled with the relative height data (the weighted difference of the input data and local estimates of the background).

The file name is set from the command line with -othrfile=file_name. If not specified, the relative height data are written to a file with the same name as the input data stack but with ".stk_thr" appended. Use "none" for file_name if you do not want the relative height data.

Return to the list of parameters


ObinFile

This parameter sets the name of the image file which will be filled with the binary images resulting from applying a threshold to the relative height data.

The file name is set from the command line with -obinfile=file_name. If not specified, the binary images are written to a file with the same name as the input data stack but with ".stk_bin" appended. Use "none" for file_name if you do not want the binary image data.

Return to the list of parameters