SlicerCaseIterator is a scripted module extension for 3D slicer. It's purpose is to streamline the segmentation of image datasets by handling loading and saving.
The input for SlicerCaseIterator is a csv-file containing the file locations of the images and/or labelmaps that have to be segmented. The first row should be a header row, with each subsequent row representing one case.
By providing column names in the module interface, columns containing (absolute or relative) paths to the images/labelmaps. The main image is loaded last and set as background image.
Additional images/labelmaps can be loaded by specifying the respective column names as a
comma separated list in the Additional images Column and Additional masks Column, with the
last image specified here loaded as the foreground image (in all 3 slice viewers).
If you already processed some part of the batch or need to start at a specific case, you can
do so by specifying the number at the Start postion parameter (with 1 representing the first case).
When input data is valid, press Start Batch and start segmenting!
When a batch is loaded, the users can navigate between cases using the Previous Case and Next Case
buttons that are then visible on the module interface.
In addition to the buttons, navigation also be controlled using 2 keyboard shortcuts:
Ctrl + N: Go to next caseCtrl + P: Go to previous case (in case the first case is active, nothing happens)
When the last case is selected and the user moves to the next case, the current case is closed and a message indicating the batch is done is shown (navigation is then disabled).
Exiting the navigation prior to reaching is possible using the Reset button,
which exits the navigation (the case is not saved and not closed).
On the python console SlicerCaseIterator prints information about the current case. Output can contain the following:
- The case number that is loaded. If the table contains a column
patientorID, the value of this cell for the current row is added to this message, e.g.Loading next patient (3/5): breast1... - An info messages when the case is closed.
- For each new file that is saved, the full path location of the new file is printed.
- Errors and warnings about invalid or missing input.
When the input is invalid (e.g. unknown column, incorrect path), error messages detailing the error are shown.
The following customization is available when processing a batch of cases:
Reader name: Any string specified here gets appended to filenames used to save labelmaps (both new labelmaps and labelmaps that were specified in the input file). This can help to prevent inadvertently overwriting files and to keep track of who made the labelmaps.Go to Editor: Check this to automatically switch to the editor module whenever a new case is loaded.Save loaded masks: Check this to enable resaving of any labelmaps that were specified in the input file and loaded (with the optional change in filename).Save new masks: Check this to enable saving of any newly added labelmaps and/or labelmaps that were not loaded by SlicerCaseIterator. Again, with to optional suffix specified inReader name.
When both Save loaded masks and Save new masks are unchecked, nothing is saved, and SlicerCaseIterator will
only show you the cases. N.B. any newly added labelmaps and changes are discarded when the user switches
to a different case!
The format of the csv file is pretty simple but needs to be customized for the layout of your data. Here's a simple example to generate a csv file for a typical layout, such as /data/Examples/pat1/pat1.nii.gz and /data/Examples/pat1/pat1-label.nii.gz.
pathFormat = '/data/Examples/pat%d'
imageFormat = 'pat%d.nii.gz'
maskFormat = 'pat%d-label.nii.gz'
with open('/tmp/chd.csv', 'w') as fp:
# write header row
fp.write("path,image,mask\n")
# write a line for each study
for patIndex in range(19):
fp.write("%s,%s,%s\n" % (pathFormat % patIndex, imageFormat % patIndex, maskFormat % patIndex))