Press Iso-Data to create the Iso-Data image(s), from the original image set, using the Iso-data algorithm. Clusters will be put in order according to decreasing # pixels in a cluster (1 - largest cluster). The regions of interest for each cluster will be saved to files for all repeated states, for the last 8 iterations, or for the last few iterations before the objective function stabilizes. The signatures for each cluster will be saved in a signature file. The signature with the minimum solution will have min appended to it. The iterations will converge if the iterations have the same solution repeated or the objective function stabilizes. The minimum solution is the iteration with the smallest objective function of the repeated solutions or the last iteration if the objective function stabilizes.

Iso-Data Results

	Select Binary to display the Iso-data results as a set of binary images, one for each cluster for Iso-Data.  Fuzzy Iso-Data will create a set of fractional images, one for each cluster.
	Select Theme Map to display the Iso-data results as a single image, with the pixel value corresponding to the cluster number.

Iso-Data Method

	Select Distance-1 to use method 1 for performing the splitting step in each iteration. This is the traditional splitting method where a constant is used to create the new cluster.
	Select Distance-2 to use method 2 for performing the splitting step in each iteration. This is a modified splitting method, where the new clusters are created from the original cluster split by comparing each pixel to the mean of the cluster center.
	Select Projection to use  method 1, but the projection of one cluster center signature to another is used in place of distance.
	Select Gram-Schmidt to use method 1, but eigenanalysis is performed on the cluster center signatures, in place of comparing distances.

Fuzzy Iso-Data

Press Fuzzy Iso-Data to create the Fuzzy Iso-Data image(s), from the original image set, using the Fuzzy Iso-data algorithm. Clusters will be put in order according to decreasing # pixels in a cluster (1 - largest cluster). The regions of interest for each cluster will be saved to files for all repeated states, for the last 8 iterations, or for the last few iterations before the objective function stabilizes. The signatures for each cluster will be saved in a signature file. The signature with the minimum solution will have min appended to it. The iterations will converge if the iterations have the same solution repeated or the objective function stabilizes. The minimum solution is the iteration with the smallest objective function of the repeated solutions or the last iteration if the objective function stabilizes.

Fuzzy Iso-Data Method

	Select 1 to use method 1 for performing the splitting step in each iteration. This is the traditional splitting method where a constant is used to create the new cluster.
	Select 2 to use method 2 for performing the splitting step in each iteration. This is a modified splitting method, where the new clusters are created from the original cluster split by comparing each pixel to the mean of the cluster center.

# Start Clusters

Enter the number of clusters to define at the start of the iterations in # Start Clusters.

# Final Clusters

Enter the number of clusters that are expected from the data set in # Final Clusters.

Minimum # of Connected Pixels

Enter the minimum number of pixels connected together for the group of pixels to be included in the cluster assignment in Min Connect Size.

Minimum Pixels per Cluster

Enter the minimum number of pixels acceptable in each cluster in Min Pixels/Cluster.

Standard Deviation Parameter

Enter the maximum standard deviation a cluster may have before is splits in Std Dev.

Press SD Roi to use the region of interest to calculate the Standard Deviation parameter.

Note

The region of interest will be saved with _i0 for an extension.  All previous regions of interest used for the Standard Deviation parameter or Lumping parameter will be ignored.

Lumping Parameter

Enter the minimum distance cluster centers may be before lumping two clusters together in Lumping.

Press L Roi tp use all selected regions of interest to calculate the optimum Lumping parameter. The Standard Deviation parameter may also be recalculated if the region of interest provides a smaller standard deviation.

Note

The regions of interest will be saved with _i# for an extension. A signature file will be saved with all the regions of interest

Maximum Pairs to Lump

Enter the number of pairs to lump together at a time in Max Paris to Lump.

Maximum # Iterations

Enter the maximum number of iterations to go through, unless a repeated pattern of changes occur, in Max Iterations.

Cluster Center Constant (Method 1)

Enter a distance parameter defining how far to go from the cluster center, to define a new cluster in Cluster Center Con. This changes the element in the cluster that caused the split.

M

Enter the exponent to be used in the Fuzzy algorithm in M.

Standard Deviation Multiplication Factor

Enter a factor to multiply the average standard deviation of signature in a file to calculate the standard deviation parameter when using the Signature button in Std Dev X.

Lumping Multiplication Factor

Enter a factor to multiply the distance between the first two signatures in a file to calculate the lumping parameter when using the Signature button in Lumping X.

Window Size

Select the Window Size, X x X, to use to test for spatial connectivity for a cluster assignment.

Minimum # Connected Pixels

Enter the number of pixels, with the same cluster assignment, connected within the window size for the center pixel to be included in a cluster assignment in # Pix Conn.

Parameters from Signature

Press Parameters from Signature to calculate the Lumping and Standard Deviation Parameters from the currently loaded image's signature file. The processes with the closest distance between them will be used for the Lumping Parameter for distance methods. The processes with the largest projection onto each other will be used for the Lumping parameter for the projection method. The processes with the maximum projection onto each other, using eigenanalysis, will be used for the gram-schmidt method. The desired process will be used for the Standard Deviation parameter.

Identify

Press Identify to identify each iso-data cluster to match a process in a signature file or eigen image. A region of interest is saved for each identified process, which may be from multiple iso-data clusters.

Angle 1

Enter the angle used to identify a final iso-data cluster as a process from a signature file in Angle 1.

Angle 2

Enter the angle used to identify a final iso-data cluster as a process from an eigen image in Angle 2.

Note

This is used only when Angle 1 fails and should be larger than Angle 1.

Enter the minimum number of points a final iso-data cluster may have to be considered for a process using Angle 1, in # Pts. Not used for Angle 2.

Minimum Percentage

Enter the minimum percentage of pixels from a final iso-data cluster that overlap an eigen image using Angle 2 in %. Not used for Angle 1.

# Standard Deviations

Enter a comma separated list of the number of standard deviations a final iso-data cluster's magnitude may deviate from the process in the signature file in # Std Dev.

Erode-4 Iterations

Enter a comma separated list of Erode-4 iterations to be performed on the identified process' region of interest. in Ero4

Erase-4 Size

Enter a comma separated list of the size of 4-connected objects, in the identified process' region of interest, that can be erased, in Era4.

Maximum Standard Deviation Change

Enter the maximum percentage the Standard Deviation parameter may change between locations for Iso-Data in Std Dev Change. If the parameter changes more, the Standard Deviation and Lumping parameters will be reduced by the value of Theta change and Iso-Data will be run again.

Note

If selected processes could not be identified, Iso-Data will be rerun also.

Enter the percentage to multiply the Standard Deviation and Lumping parameters before running Iso-Data again in Theta Change. Iso-Data is run again if the Standard Deviation parameter changes by more than the Maximum Standard Deviation Change

Standard

Press Standard to standardize the displayed Iso-Data image by remapping the cluster values to new cluster values, based on the distance between each cluster signature and two selected region of interest signatures. The user enters two new cluster values for the two selected regions of interest. All of the clusters in the displayed Iso-Data image will be remapped between the two new cluster values, based on their distance or angle to the two defined regions of interest.

	Select Distance to use the distance between cluster signatures to determine the new cluster values.
	Select Angle to use the angle between cluster signatures to determine the new cluster values

Standard Remap Values

	Region/Signature 1 Press Region/Signature 1 to define the first standard signature from an Iso-Data cluster, a displayed region of interest, a selected region of interest file, or a selected signature file. Note Enter a value, to remap the cluster value to, in remap to.
	Region/Signature 2 Press Region/Signature 2 to define the second standard signature from an Iso-Data cluster, a displayed region of interest, a selected region of interest file, or a selected signature file. Note Enter a value, to remap the cluster value to, in remap to.
	Signature from Select Iso-Data Cluster to use the selected Iso-Data cluster signature as the standard signature. Select Region of Interest to use the displayed region of interest, or the selected region of interest file, to calculate the standard signature. Select Signature to use the desired process from the selected signature file as the standard signature.  See Make Signature in the Roi/Voi Dialog on how to make a signature of a process from another slice, if needed.

Repeat

Press Repeat to perform substance on other images.

Note

Previous results of repeat processing need to be deleted manually for images saved in Dicom format.

Cancel

Press Cancel to remove the Substance dialog.

Help

Press Help to display help for the Substance dialog.