S - Troubleshooting

Enhanced CellClassifier starts, but the main window remains grey:

Enhanced CellClassifier remains in the "Save-Mode". In order to start, Enhanced CellClassifier is missing data from a CellProfiler output file. Use >> File >> Merge and split data, to bring the output file into the right format. Choose either >> File >> New Experiment or >> Settings >> Modify Settings followed by the "Browse" button of the "Load data path" and select the _handles.mat file of the newly split data.

Enhanced CellClassifier starts, the main window remains black:

No original images are available. Choose either >>à File >> New Experiment or >> Settings >> Modify Settings. Use the "Browse" button of the "Load original images path" and select the appropriate folder. Alternatively press "LUCKY" for an automatic selection.

Enhanced CellClassifier and displays images but many functions are disabled

Possiblity 1: No Segmented images are available. Either no segmented images exist or the path is incorrect. For selection of objects on the images, CC needs Segmented Images (=LabelMatrixImages). These images indicate the borders (segmentation) of the objects. You need to create these images with CellProfiler (module: ConvertToImage, option 'uint16', and module: SaveImages as 'mat' file).
Select the path to the segmented images either in >> File >> New Experiment or >> Settings >> Modify Settings.
Possiblity 2: Segmented images exist in the correct folder but cannot be recognized by Enhanced CellClassifier. The filenames for the segmented images are generated by replacing string 1 by string 2 in an image-filename (for instance, replacing '.tif' by '.mat'). You can adjust these strings in the "Define Settings" window.

Possibility 3: No data are selected. You can select the data that will be used for classification in the "Define Settings" window (>> Settings, >> Modify Settings). Press the "+" button to add a new measurement. Make sure, that only objects which necessarily have the same number in each image are combined (for instance 'Nuclei' and 'Cells', when each cell is generated from a nucleus).

The libsvm-path is found, but the program does not seem to be there, and the SVM-window does not open

LIBSVM is incorporated in the Enhanced CellClassifier program. It needs .mex files (C++ files which are Matlab-executable) to work. .mex files might be different for different environments (32-bit windows, 64-bit windows, Macintosh, etc.). Enhanced CellClassifier checks, whether the respective files for the current environment are there, if not, the SVM-window does not open.

The .mex files for 32-bit windows, 64-bit windows, and MacIntel are already there. If you are working in a different environment, you might need to make these files yourself. Please read the instructions in the "readme" file, in the LIBSVM-folder in the Enhanced CellClassifier program and act accordingly.

For Macintosh the following strategy worked (thanks Micha):

• install gcc-compiler (Xcode). Should be on the Macintosh installation disk or can be downloaded from Apple
• in folder libsvm change in file "MakeFile" one line:
  the 3^rd line gives the Matlab-path, this needs to be adapted to be the path where Matlab ist installed
• execute the command make

If all this does not work, contact LIBSVM (or us) for technical support. http://www.csie.ntu.edu.tw/~cjlin/libsvm/