An interactive prototype to help correct the cell labels predicted by state-of-the-art image segmentation algorithm.

• change the contrast of the image
• correct the segmentation by skeleton
• When saving, export to a file type readable by amria/vtk

• Three views.
• The color image of predicted segmentation is overlayed on the origin cube data. Different colors encode different labels.
• resize the image view with ctrl + mouse wheel and use middle mouse to drag the image
• scrolling through z planes with mouse wheel
• quick location with middle mouse click + ctrl
• permit adjustment of opacity of segmentation overlay (there is a trackbar where you can adjust the opacity of label opacity%)
• Membrane Edition
• permit control over appearance of outlines. 1. For the selected ID, 1.1 permit the user to toggle whether the outline is visible (press 'q' to toggle or toggle off), 1.2 provide control over outline opacity or thickness.(there is a trackbar where you can adjust the selected cell's outline opacity%) 2. For the all segments, 2.1 permit the user to toggle whether the outline is visible (there is also a trackbar where you can adjust the selected cell's outline opacity%) (press 'w' to toggle or toggle off)
• Merge Cells left mouse click selects which segmentation ID will receive additional segments (updated as foreground), right click specifies which region (segment ID) should be added to the outlined, selected ID.
• Extend outline of selected segmentation ID to all z planes
• Permit operations on arbitrary “selections” (areas within a single Z-plane). (detailed description can be seen below)
• Split cells
• Clear small fragments by pressing 'ctrl + c'
• multiple-step undo.
• save the current annotation: press 'ctrl + s' and it will saved in annotation_history/ (timestamp is used to distinguish between different versions.)

0. Use the drop-down menu to go to "membrane edition" mode.
1. Hold on left mouse as pencil to add membrane / Ctrl + hold on left mose as eraser to remove membrane.
2. Ctrl + D flood-fill

0. Use the drop-down menu to go to "merge (global)" mode.
1. Left mouse click to select host ID. After selection, pixels of the same label will be highlighted. And foreground label will be updated.
2. Right mouse click to select other ID to be fused with the host.
3. Press 'ctrl + b' can toggle back the previous "adding"

Special note:

• You can click right mouse continuously to add the current cell to the "receive" cell.
• You can also press 'b' continuously to restore the'added' segmentation to its original ID step by step. (MAXIMUM 8 steps will be stored, means you can only recover 8 steps.)

1. belong to a particular user-specified ID

   1.1. **Use the drop-down menu to go to "edit one ID (slice)" mode. **

   1.2. press left mouse + ctrl to select specified ID which will be relabeld(it will be highlighted and the ID color will show in background)

   1.3. Hold on right mouse to draw an arbitrary region

   1.4. after right mouse up, selected region with particular ID will be highlighted

   1.5. change label by

    * **Left Mouse click + ctrl** on other region
   

2. all region

   2.1. **Use the drop-down menu to go to "edit all IDs (slice)" mode. **

   2.2. Hold on right mouse to draw an arbitrary region

   2.3. after right mouse up, all selected region will be highlighted

   2.4. change label by

    * **Left Mouse click + ctrl** on other region
   

special note:

• labels won't be propagated through z-plane
• when you are drawing, keep the mouse pressed

0. Press 'v' to go to splitting mode. (or use the drop-down menu)
1. hold on shift and move the mouse (keep it pressed) to draw an cut (there is a slider that you can adjust the width of the line you draw)
2. the program will test whether the cell is splitted (across z index), if it's splitted, will assign the splitted part a new id/color.

1. press 'ctrl + b'

Special note:

• MAXIMUM 8 steps will be stored, means you can only recover 8 steps.

1. permit adjustment of opacity of segmentation overlay

   • there is a trackbar where you can adjust the opacity of label opacity%

2. permit control over appearance of outlines.

   2.1. For the selected ID,

    2.1.1 permit the user to toggle whether the outline is visible      
        * press 'q' to toggle or toggle off 
    2.1.2 provide control over outline opacity.
        * there is a trackbar where you can adjust the opacity of selected cell
   

   2.2. For the all segments,

    2.2.1 permit the user to toggle whether the outline is visible 
        *press 'w' to toggle or toggle off
    2.2.2 provide control over outline opacity.
        * there is a trackbar where you can adjust the opacity of all cells
   

1.mouse wheel

2.middle mouse click + ctrl to quick locate

1. press 'ctrl + s'

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes.

• python2.7
• opencv2
• numpy
• pyqt4

console -> open an ipython console

in ipython:

• import sys
• sys.path.append('path to the code folder')
• import annotate_gui_connectomics
• annotate_gui_connectomics.main()

in GUI:

• select sem_file, seg_file and output_folder in file dialog
• close file dialog

• Shuqi Wang - Initial work