Watershed workflow

documentation/watershed/watershed.md

@@ -0,0 +1,118 @@
+---
+layout: documentation
+title: Watershed
+tagline: Interactive Segmentation
+category: "Documentation"
+group: "workflow-documentation"
+weight: 4
+---
+# Watershed
+
+## How it works, what it can and cannot do
+
+"It is important to clarify that
+small changes in the seeds can have a huge impact on the outcome of the watershed.
+Therefore, it is important to have an interactive graphical interface which allows the user
+to directly see the new outcome of any changes made on the seeds. The topic of this
+thesis is the expansion of this software with a new watershed workflow.
+The workflow needs to be able to handle the following use cases where the following data
+can be supplied:
+1.) seeds (nuclei) and/or 2.) boundaries
+
+• Seeded Watershed:
+1. only seeds: pointless to do a seeded watershed with seeds only.
+2. boundaries and seeds: use a seeded watershed on boundaries with the supplied seeds. The seeds can be changed, removed or new seeds can be drawn manually. 
+3. only boundaries: generate seeds out of the boundaries and label them. Carry
+out the use case where the seeds and the boundaries are supplied.
+
+• Unseeded Watershed should still be possible when only using the boundaries.
+
+In ilastik a workflow is made up of several applets that are connected with each other.
+The seeds applet contains the functionality of generating seeds and sets the watershed
+algorithm to seeded or unseeded. It is possible to manipulate, create or delete seeds
+manually in the watershed applet. The watersheds of the boundaries with the given seeds
+can be computed and visualized. Any change to the seeds and their impacts on the
+watersheds are visible immediately. In the end the resulting data can be exported.
+" **cited from [chapter 1. Introduction] Andreas Haller, Master Thesis, "Interactive Watershed Based Segmentation For Biological Images", March 2017, Heidelberg University**
+
+
+
+## Generate Seeds 
+"In the seeds applet the user can view all provided inputs, generate new seeds constructing
+them on the given boundaries or decide whether the following watershed should be carried
+out unseeded independently of given/generated seeds.
+
+**Usage**
+
+• Unseeded:
+if checked, the watershed algorithm will be an unseeded Union Find algorithm
+
+• Smoothing:
+a selection of smoothing methods can be chosen
+
+• Sigma:
+the smoothing parameter that describes the intensity of smoothing
+
+• Compute Method:
+a selection of methods that can be chosen to compute the generated seeds
+
+• Generate Seeds Button:
+needs to be clicked to generate seeds and override any provided seeds with generated
+ones
+
+Seeds are generated by:
+1. smoothing the boundaries with the given smoothing method and the given sigma
+2. computing seeds on the basis of the smoothed image with the selected compute
+method
+
+<a href="fig/seedsApplet.png" data-toggle="lightbox"><img src="fig/seedsApplet.png" class="img-responsive" /></a>
+" **cited from [chapter 3.2.2 Seeds Applet] Andreas Haller, Master Thesis, "Interactive Watershed Based Segmentation For Biological Images", March 2017, Heidelberg University**
+
+If the image of the supplied seeds is binary, this image is automatically labeled where 0 is background.
+
+
+## Interactive Watershed Segmentation with manipulation of Seeds 
+
+"In the watershed applet the user can execute a watershed algorithm. This algorithm
+depends on the watershed method that is passed on by the seeds applet. Additionally,
+the user can manually modify, delete and create new seeds. It is possible to reset any
+changes to the supplied seeds. The impact of these changes on the results of the watershed
+algorithm can be directly seen.
+
+• Neighbors:
+specify the number of pixel that are taken into account
+* direct: 2D means 4 neighbors &
+	3D means 8 neighbors
+* indirect: 2D means 6 neighbors &
+	3D means 26 neighbors
+
+• Show “Corrected Seeds” pixel value:
+shows the pixel value of the “Corrected Seeds” layer where the cursor points to
+
+• Reset “Corrected Seeds” to “Seeds”:
+remove all changes on the seeds that are done manually (that are displayed in the
+“Corrected Seeds” layer), so that they are identical to the “Seeds” layer, which
+displays the originally input seeds
+
+• Drawing/deleting/changing labels:
+– drawing:
+select a given seed, click on the brush icon and draw into the image
+– deleting:
+click on the red circle with a cross inside to delete the label from the label list
+and its associated pixels from the ”Corrected Seeds“ image
+– clearing:
+right click on a label and click ”clear Label“ to remove all pixels with this value
+from the seeds image. This label isn’t removed from the label list and can be
+used later on
+– ”Add Seed“ button:
+add an additional label (limit: 254 labels)
+– rubber:
+reset everything under the cursor to 0
+
+• run seeded watershed:
+execute the watershed algorithm on the boundaries. If seeded, then use the ”Corrected Seeds“ as seeds
+
+<a href="fig/watershedApplet.png" data-toggle="lightbox"><img src="fig/watershedApplet.png" class="img-responsive" /></a>
+<a href="fig/watershedApplet_geklickt.png" data-toggle="lightbox"><img src="fig/watershedApplet_geklickt.png" class="img-responsive" /></a>
+" **cited from [chapter 3.2.3 Watershed Applet] Andreas Haller, Master Thesis, "Interactive Watershed Based Segmentation For Biological Images", March 2017, Heidelberg University**
+