Welcome to the modified version of Nih-Image 1.61 (Image 1.61Hull.68k)

Nih-Image is probably the best public domain program available for working on nuclear morphometry with Macintosh platforms. 
One problem, however, is the manipulation and processing of data generated by the program for  both 
descriptive and inferential purposes.
In fact, if you want to use the shape descriptors data as covariates in survival models, predictive models or whatever else, you usually need:
a) to export the measurements for each image as a Tab delimited text file
b) compute descriptive stadistics (mean, sd, skeewnes, etc) with an external application
c) accumulate by hand these value in another archive for further processing

As this problem arose at my Department, where investigation on nuclear morphometry in different urological
tumors is ongoing, I decided to modify and recompile Version 1.61 to include the following features:

1) A new macro named NewDataFile('FullPath') to create a sequential access read/write text file
2) A new macro named ComputeData('String') to compute descriptive statistics, which include the Mean, Standard Deviation Skeewness and Curtosis for the selected measurements and three additional shape descriptors,
represented by  the Form Factor, the AspectoRatio (MajorAxis to MinorAxis ratio) and Compactness.
The generated file may be exported to other applications that accept Tab delimited text files (Excel, SPSS, Survival, 
Fox Pro, etc.) as input data.
3) A new option to compute the Convex Hull of particles, which is very fast and efficient.
If User1 and User2 are selected, the Convex Area and the Normal Area to Convex Area ratio (Solidity) are computed and displayed in the Results windows. 

To use the Convex Hull option follow this steps:

1) Select the Convex Area item from the Measurements options dialog box. 
2) Check the User1 and User2 item to compute the Convex Area and Solidity, which are displayed in the Results window. They are also computed with the new ComputeData macro.
3) You may measure the Convex Hull of individual particles with the Wand Tool by checking the
Wand Auto-measure item.
4) If you check the Outline Particle option from the Particle Analysis Options dialog box, the boundaries of the
convex polygon will be traced. If unchecked, all the convex area will be filled with the background color


The ComputeData macro has a string parameter to specify the type of analysis to be performed and may include any combination of the following symbols:

'Mean' to compute the average value.
 
'Sdev' to compute the standard deviation.

'Skeew' to compute the skewness coefficient.

'Curtosis' to compute the curtosis coefficient.

'Labels' to add labels to the file created with the NewDataFile macro. It is used to identify the different 
measurements (up to 288) generated by the macro. Labels are inserted at the first line of the data file.

'FormFactor' to compute the Form Factor.

'AspectRatio' to compute the MaxDiameter to MinDiameter ratio (Ellipticity).

'Compactness' to compute compactness (sqrt(4/pi) * Area)/MaxDiameter.

'Standard' to compute the above statistics for the whole measurements.

'SmallTen' to compute the above statistics for the ten smaller nuclear areas.

'LargeTen' to compute the above statistics for the ten larger nuclear areas.

The ComputeData macro sequentially appends all this values to the file created with the NewDataFile macro, which can be exported to any other application as a Tab delimited text file for further processing.

Here is a short example on how to use this macro for automated particle analysis, which can be changed
to fit to your needs. Other macros are included in the ComputeData macro file.

macro 'DATAPROCESS';

var
i, nPictures:integer;

BEGIN
 nPictures:= 200;
 NewDataFile('Development:MyDataFile'); {specify the full path}
 for i:= 1 to nPictures do
 begin
   SelectPic(i); {it could also be a Stack or Files stored in your Hard disk. See the ComputeData macro}
   SetDensitySlice(224,254); {add additional segmentation commands as needed}
   SetOptions('AreaLengthMajorMinorAngle'); {add any other measurements you want}
   AdjustAreas(true);
   AnalyzeParticles('OutlineIgnoreIncludeReset');
   ComputeData('MeanSdevSkeewCurtosisLabelsStandardSmallTenLargeTenFormFactorAspectRatio');
  {Computes and sequentially adds these values to the previously created data file}
   RevertToSaved;
 end ;
END;

If you use the same File name in other analysis, new data are appended at the end of the file until you change the name of the file or you close the application.
The above example generated a text file with 7 * 3 * 3 columns by 200 rowswith an average of 100 nucleus per picture and an average speed of 12 seconds per picture in my old LC-Mac (with a FPU). You can't imagine how much time and trouble you may spare.

If you work with nuclear morphometry in malignant tumors and want to use the morphometric descriptors
as covariates in parametric and nonparametric survival models have a look at my Public Domain Survival614b program (you may get a copy by anonymous ftp in any Info-mac server e.g. sumex-aim.stanford.edu, currently at the _Science&Math Folder}

ÁHave Fun and Enjoy it. Comments and suggestions are wellcomed !

 Manuel Urrutia Avisrror, MD., P.h.D 
 Catedratico de Urologia                        
 Facultad de Medicina de Salamanca     
 Salamanca - SPAIN                    
 e-mail: urrutia@gugu.usal.es