We have a lot more vertical space now that we are using tabs. We should take advantage of this by increasing the vertical size of the tables.

PCampReview (Prostate CAncer Multi-Parametric REVIEW) was my idea, and I am no longer happy with it (it is too long, and confusing, and may be unnecessarily narrowing the scope).

How about just mpReview?

Any other ideas?

@rweiss42 @che85 @PeterBehringer your comments are welcome!

Need to implement checks at the following levels:

1. consistency before saving (only one label is present, label number and structure name are consistent, etc similar to as implemented in https://github.com/fedorov/PCampReview/blob/master/Util/CheckAndFixLabels.py)
2. consistency after saving (read segmentations back and check as above)
3. visual checking (save screenshot of how the label shows up at the time of saving): this one is most tricky - probably makes sense to go over one label at a time, set slice to the center of the label, turn on slice viewer annotation and save the screenshot of the slice layout for each label)

1 and 2 should be easy and quick, 3 is for later.

Following scenario:

1. User selects a input directory which has sub folders
2. mpReview interprets those directories as studies (where actually should already been done some check with feedback if it fails)
3. going to tab 2 (series selection) throws exception caused by missing RESOURCES subdirectory....

There might be a problem with the fiducial table. Slicer also crashes, when I delete the currently selected fiducial list node.

In the old UI layout, we could see the titles of the previous steps which would show the user what data directory they're in and what patient/study was selected. I'd like to add some labels (or something else? suggestions?) either globally or just in step 4 to show that data directory you are in and what study you're currently working with. Similar to how we added a label in step 2 to show what data directory you are currently in.

When mpReview study is selected and series are preselected the segmentation tab is not enabled.
First the user has to click into the series listview so that the segmentation tab is activated (enabled)

When using mpReviewPreprocesser the user should get feedback about the current progress to make sure that he/she doesn't think that the application got stuck.

Selecting more than 16 series causes a divide by zero error on line 1407. This is because the if statement on lines 1399-1406 does't catch the case when nVolumeNodes > 16.

When clicking on "View source code" the source code should be opened in a web browser

When you're warned that you're going to loose any unsaved contours on leaving step4 or 5, can we add a button to the warning pop up that calls the save method? Otherwise you have to go to completion step and click save there.

Better yet, track if things have changed since last save and only warn the user if in fact they're about to lose something. But that sounds complicated...

It would be great if mpReview could offer functionality for minimizing the need of user input.

Proposal: If directory with none mpReview compatible directory structure is found which is eligible for mpReviewPreprocesser, a prompt could be opened which shows:

• Output directory where to store the mpReview-"project"(DICOM files would automatically organized).
  mpReviewPreprocesser would be executed and input path is after execution set to the output directory of mpReviewPreprocesser

@rweiss42 @fedorov What do you think about that?

What you think about my current implementation?

I added an new workflow step and icon.

TODO:

• implementation of fiducials setting
• discarding fiducials when discarding changes

• currently targets are saved in a Targets directory next to the RESOURCES directory

Iterators, access to reconstructions, DICOM data, etc should be abstracted out into API to simplify usage and facilitate eventual substitution with a different back-end.

Since SliceAnnotations were revised in Slicer/Slicer#373 I get the following error:

Traceback (most recent call last): File "/Applications/Slicer.app/Contents/lib/Slicer-4.4/qt-scripted-modules/DataProbeLib/SliceViewAnnotations.py", line 582, in updateViewAnnotations self.sliceViewNames.append(sliceViewName) AttributeError: 'tuple' object has no attribute 'append'

I'd like to propose that when you enter into a reference image and there are no structures yet, the editor area is disabled. Once a structure is created, enable it. On delete, check to see if there are no structures left and if so, disable the editor again.

I think we should talk about usability aspects.

Some proposed improvements:

1. Having only one tab for "data selection". The current approach is a waste of space and the user needs to interact more than needed. I would leave the first tab like it is and add another splitview below the study selection list view.
2. Simplifying the segmentation tab. I got the feeling that the segmentation tab is far too full. The fiducials table somehow is lost in that layout when you don't know about that (unexperienced users)

@fedorov @rweiss42 What do you think?

When adding a new structure and using the PaintEffect/Draweffect and then going to Completion step, you will still be able to paint/draw in the viewers. This also affects switching to another module where the effects are still enabled.

We should disable them as soon as possible after finishing up.

To reproduce:

1. select study
2. go to segmentation tab
3. open collapsed buttons (so that the UI height exceeds the display)
4. go back to data selection tab

We can probably use the Editor "split merge" feature here -- include the merge frame in the GUI and integrate with the logic. When reporting form is invoked, populate with the label ID for the active label in Editor to link finding reporting.

Unfortunately, it appears that the width is set to the maximum width, which is the Structures widget in the Editor module. On 15-inch mac retina screen, module panel steals over 50% of the screen width!

The layout used there is QHBoxLayout, and it keeps getting wider with the recent modifications of the functinality. I guess it would be great not to use QHBoxLayout in this part of Editor, or use combination of icons and tooltips instead of long button labels. @pieper any other suggestions?

Sceenshot below is full width of 15-inch mac screen.

Should be slicer.cli.run(....)

In order to be able to skip every second slice while for instance segmenting the whole gland, we need to add back again the DilateEffect

The idea is to organize results of OncoQuant in the same hierarchy for different options of PK analysis as below:

└── RESOURCES
    ├── 1
    │   └── Reconstructions
    │       ├── 1.nrrd
    │       └── 1.xml
    ├── 100 (where type is SUB)
    │   ├── Reconstructions
    │   │   ├── 100.nrrd
    │   │   └── 100.xml
    │   └── Segmentations
    │        └── readerId-StructureName-datetime.nrrd
    │   ├── OncoQuant
    │       └── version-selectionofaif-MapName-datetime.nrrd

where selectionofaif is one of the following: Auto, ManualLeftArteryROI, ManualRightArteryROI, Model. DCE series should be used to run the analysis, but the results should be saved under SUB series in OncoQuant folder.

The result should be a bash or python script that takes on input the locaiton of PCampReview-style dataset, and adds OncoQuant directory with the maps. The following maps should be saved: Ktrans, ve, BAT, Rsqr, Auc, MaxSlope

Need to make sure the default LUT currently used contains all the terms needed by the the Chicago group.

We should have a dropdown or something similar to provide easy access to previously used data directories. This is especially useful for those who often has to switch between directories.

This can be saved to module specific settings.

Obviously mpReview needs to check if those directories exist and they are mpReview eligible.

@mehrtash can you take a look into this?

If I look at the header of the segmentation file, orientation/direction of the image are different from those of master volume. I do not know if this is a bug or expected behavior. It is probably an Editor issue.

Execution under Windows works for the very first time. Starting PCampReviewProcessor again, an error occurs regarding the deletion of the temporary path:

self.dataDir = os.path.join(slicer.app.temporaryPath, "PCampReviewPreprocessor")
shutil.rmtree(dataDir)

Somehow even when I execute this code manually after starting Slicer but without calling PCampReviewProcessor, Windows replies:

WindowsError: [Error 32] The process cannot access the file because it is being used by another process: u'C:/Users/CHRIST~1/AppData/Local/Temp/Slicer\PCampReviewPreprocessor\CtkDicomDatabase\ctkDICOM.sql'

use subject hierarchy to organize the data internally, as either a parallel data structure, or a replacement for the data structure currently used.

see http://www.slicer.org/slicerWiki/index.php/Documentation/Nightly/Modules/SubjectHierarchy

To reproduce create structure, paint, select eraser, try erasing

Since there can be many studies in one data directory which has to be processed, it would be very useful to give feedback to the user about the current study status.

Traceback (most recent call last):
File "/Users/peterbehringer/MyProjects/Modules/PCampReview/PCampReview.py", line 318, in setup
effectButtonFrame.addWidget(slicer.util.findChildren(editBoxFrame,'WindowLevelEffectToolButton')[0])
IndexError: list index out of range
Traceback (most recent call last):
File "/Users/peterbehringer/MyProjects/Modules/PCampReview/PCampReview.py", line 202, in onTabWidgetClicked
self.onStep3Selected()
File "/Users/peterbehringer/MyProjects/Modules/PCampReview/PCampReview.py", line 1211, in onStep3Selected
self.resetTranslate()
File "/Users/peterbehringer/MyProjects/Modules/PCampReview/PCampReview.py", line 1930, in resetTranslate
self.translateLR.value = 0
AttributeError: PCampReviewWidget instance has no attribute 'translateLR'

code from Thiele

       double rms = optimizer->GetOptimizer()->get_end_error();
       double SSerr = rms*rms*vectorVoxel.GetSize();
       double sumSquared = 0.0;
       double sum = 0.0;
       double sumTest = 0.0;
       double ssSum = 0.0;
       for (unsigned int i=0; i < bValuesSelected; ++i){
           sumTest += (imageValuesPtr[i]-fittedValuesPtr[i])*(imageValuesPtr[i]-fittedValuesPtr[i]);
       }
       for (unsigned int i=0; i < vectorVoxel.GetSize(); ++i){
         sum += vectorVoxel[i];
         sumSquared += (vectorVoxel[i]*vectorVoxel[i]);
         ssSum += (vectorVoxel[i]-fittedVoxel[i])*(vectorVoxel[i]-fittedVoxel[i]);
       }
       double SStot = sumSquared - sum*sum/(double)vectorVoxel.GetSize();
       double rSquared = 1.0 - (SSerr / SStot);
       double chiSquaredNorm = sumTest/((double) bValuesSelected-4.0);//This number depends on the number of fitted parameters!
       double chiSquared = 2.0* sqrt(chiSquaredNorm);
       double chiSquaredNormAll= ssSum/11.0;
       double chiSquaredAll= 2.0* sqrt(chiSquaredNormAll);
       rsqrIt.Set(rSquared);
       ssIt.Set(ssSum);
       ssnob0It.Set(sumTest);
       ChiIt.Set(chiSquaredAll);
       Chinob0It.Set(chiSquared);
     }

Simplify/clean up the code without changing any of the functionality.

Users should be able to see processed data of other users which is saved under another username. To ensure, that the directory is readonly, we can disable ui elements for that moment.

To be discussed: Where and how to provide this to the user (dropdown?)

In order to be able to use it from other modules without duplication of code

Using volume clip would improve timing for creating labels