The Pipeline Window

Image Analyst MKII uses flowchart-based pipelines to automate all aspects of image processing. Built-in pipelines (see Pipelines main menu) support a range of biological applications from basic image processing maneuvers to complete assay analysis. The behavior of built-in pipelines can be controlled through the parameter bar in a similar way to any other image processing function, and the flowchart structure of the pipeline remains hidden for basic use. Any of the built-in pipelines can be opened in the Pipeline Window, modified and saved under a new name or custom pipelines can be created from scratch. This page explains how to create new or modify existing pipelines working with the flowchart structure. For basic pipeline usage please see the Parametering Image Processing Functions and Pipelines page. During basic pipeline usage only select (master) pipeline parameters can be modified, but not the pipeline structure or the individual function parameters within the pipeline.

The Pipeline feature provides a flowchart-like automation of image processing.

Easy, drag-and-drop style editing, no programming skills required
Fast, parallel processing
Processing of all stage positions and whole or partial microplates
Modify existing built-in pipelines or build new ones from scratch.
Select key function parameters as pipeline master parameters for quick access from the parameter bar.

To access the Pipeline window:

To create a new blank pipeline, in the Main menu use the PipelineDesigner/'New Pipeline window' menu item, or press the main toolbar button if no pipeline has been activated on the parameter bar.
Use the main toolbar button or Tools/'Show Selected Pipeline' to show the Pipeline Window corresponding to the pipeline activated on the parameter bar. Note: any modification to the built-in pipeline will be automatically saved, therefore it is advised to resave the pipeline using the save as button of the Pipeline Window tool bar. The Pipeline Window prompts the user to do this, but this is not mandatory.
Pipelines can be loaded into a blank Pipeline Window by pressing the open toolbar button in the Pipeline Window, or by dropping *.ips files to the main window.
To find an already opened Pipeline window use the PipelineDesigner/Show menu item.
Many pipeline windows can be open in the same time, but only one menu-selected pipeline will be shown on the parameter bar of the main window.
It is suggested to use a secondary display for Pipeline Windows.

To run a pipeline in a Pipeline Window (not selected in the main window parameter bar) :

on the Pipeline Window toolbar will execute the pipeline on the top Image Window. The same button in the main toolbar will repeat the pipeline from the last used Pipeline Window if no pipeline is selected in the main parameter bar.
on the Pipeline Window toolbar allows automated loading of the image data followed by pipeline execution. The pull down menu of this button allows processing single, multiple or all stage positions in a recording. To automatically load recordings the last opened Multi-Dimensional Open dialog is used, unless tags are set for dialogs and a tag is set in the 'Input Image' functions of the pipeline by selecting e.g. "[Multi-Dimensional Open]#1" as Image Window parameter, where 1 is the tag. In this case the Multi-Dimensional Open dialog with the corresponding tag is used. When [Actual window] set as Image Window parameter, the pipeline always starts processing with the topmost Image Window after performing loading of image data.
Use the PipelineDesigner/Run... main menu item
To run a pipeline selected in the main parameter bar, see instructions here.
Pipelines can be also run by the Load and Run Pipeline function.


Toolbar 2. Select function to add with the button, or use the add selected function
Workspace 3. drop the selected function into an empty area. Then, drop function onto its input. Copy/ Paste/ Rename functions by right-click
Use the button to lock the workspace, so function parameters can be set without accidentally altering the structure of the flow chart.	Red functions create/or return with new image windows, so multiple red functions can be the output of any function. Black functions modify images so only one black function can be the output of an other function.	Green functions result non-image data, therefore not other function can be the output of these functions.	Parameters: identical to the parameter bar of the main window. 4. Click a function in the Workspace, then set image processing parameters here.

		1. Always start with an Input image (use to add)		5. Run pipeline

Modifying existing pipelines:

When an existing pipeline is opened in the Pipeline Window, the editing of the pipeline structure is initially locked by the toolbar button. If the pipeline has master parameters, these can be edited in the function parameters section of the Pipeline Window on the right, while the pipeline structure is locked. Use the pull down menu of the toolbar button to adjust the level of the lock, or press the toolbar button to unlock the pipeline.

Select a pipeline in the Pipelines menu.
Press the main toolbar button to show the Pipeline Window.
Press the Pipeline Window toolbar button to unlock the pipeline.
Follow the prompt to save the pipeline in the My Pipelines folder. Note: the location of the My Pipelines folder can be changed in the 'Set Folder Locations and Default Open Method' dialog, that is accessible from the File main menu. Contents of the My Pipelines folder are also shown in the Pipelines menu.
To rename the pipeline (the pipeline caption appearing in the Pipelines main menu), press the toolbar button, and select the 'Pipeline Description' tab in the bottom of the window. Change the caption. Press the toolbar button again to hide the tabs in the bottom.
Click any function in the Workspace of the Pipeline Window to edit parameters on the left.
Follow the rules below to add or remove functions.
To modify master parameters (the parameters, not their values) press the toolbar button and work on the 'Pipeline Parameters' tab.

Pipeline editing rules:

Always start with an input image. Multiple input image can be used if the name (caption) of each image window is set in the Input functions, or using tagged Multi-Dimensional Open dialogs. In the latter case set the tag of the corresponding dialog in the 'Input Image' functions of the pipeline by selecting e.g. "[Multi-Dimensional Open]#1" as Image Window parameter, where 1 is the tag.
See available functions in the Functions Glossary.
Place new functions in an empty area, then drag and drop onto their inputs.
Functions with multiple inputs have to be dropped onto their inputs ordered.
Inputs and outputs can be reordered or deleted by clicking them in the parameter list on the right, and then clicking on the button appearing on the right:

Reorder Inputs

Use right click to copy, paste or rename functions. Functions can be arbitrarily renamed.
Red functions are inputs, seek for certain channel or create a new image. Multiple red functions can be dropped on any (black or red) functions.
In place operations are shown in black, these alter the image data, so only one black function can be the output of any function.
Functions that do not produce image as outputs are shown in green, so these are dead ends.
The order of execution is unpredictable because of parallel processing. Use Wait for all inputs if processing is needed to be synchronized at some point, e.g. when defining and using reference images.

Rules of pipeline running:

The pipeline processing starts from the input image(s) and performed parallel where multiple outputs (red functions, like Copy or Get Linked Channel) are present.
The number of image processing threads that are actually parallel processed can be set in the Preferences dialog.
If a function requires multiple inputs, it will wait until all inputs are ready. The processing continues on other branches meanwhile.
The flowchart is not fool-proof, the user have to make sure that the processing is feasible, e.g. the inputs of a function are not dependent on its own output, so there are no loopbacks in the diagram.
In the Pipeline not linked images can be processed together, therefore the user has to ensure that co-processed images have the same x,y,t sizes. Mismatched image size will cause the pipeline to stall.
Pipeline running can be stopped by pressing ESC or using the button.
If the pipeline fails to run completely press the button to reset it. The incomplete run is indicated by the pipeline progress indicator in the main window status bar, or by the incomplete graying of functions in the pipeline flowchart. The currently processed function is highlighted in green.

Using pipeline master parameters:

The master parameters are the parameters shown in the main parameter bar of Image Analyst MKII for the pipeline and control one or more function parameters within the pipeline. Master parameters are used to provide 'biological decoding' of pipelines functions for basic or routine use of the pipeline. See tutorial here: Adding master parameters to a pipeline

Tutorials:

Overlay, align and filter multi-channel image sequences and export as avi
Calculation of fluorescence intensity ratio plots after channel and series alignment and background subtraction, using automatic ROI drawing

Adding master parameters to a pipeline

Notes:

The Main Window and Image windows are responding to user input during pipeline processing, and the currently processed image windows are not protected for user interference. Therefore it is highly advised to avoid any user input during pipeline processing.
The Pipeline does not offer flow control or substitution of image processing parameters with variables (with the exception of file naming). For a full powered scripting use the Mathematica through Mathlink feature. Through Mathematica all function parameters can be fully controlled