ZEISS Knowledge Base

Introduction

This guide explains you how to detect nuclei and cells using the pipeline with the Machine Learning Segmenter operation. Machine learning (ML) is a branch of Artificial intelligence (AI) in which, based on the training dataset that are first provided, the computer develops its own logic for answering future questions. The key concept of machine learning is to produce accurate predictions on new unseen data after being trained on a finite learning dataset.

Activating the Sample Pipeline

In the Shortcut Toolbar, click Analysis Panel .
In the Sample Pipelines list, double-click the Detect Nuclei And Cells Using Machine Learning Pipeline.
If you have activated a pipeline, it will be replaced by the new one.

You can open the appropriate How to guide. By hovering over a pipeline, this button is displayed. When clicking on it, the option Open How to appears.

The following operations are part of the pipeline.

This operation allows to select the region of interest (ROI). ROI defines the dataset subarea that will be processed and analyzed by the pipeline.

Parameter		Description
ROI		Sets the processing and analysis target space.
–	Current View	The selected Z plane and the viewer area are processed.
–	Current Plane	The selected Z plane is processed (XY).
–	Current Time Point	The selected time point is processed (XYZ).
–	Current Image Set	The complete dataset (XYZ and time) is processed.
–	Custom	Allows to mix the previous methods. Note: Use the Custom option during the pipeline setting and testing. Set a sub volume (XY, Planes, Time Points, channels) of your dataset on which perform the trial. This will speed up the setting process.
Channels		Sets the processing and analysis target channels. Selecting a single channel, all the operations in the pipeline will be forced to use it.
Scaling		Sets the scaling of the dataset, which reduces it size. The measurements will not be modified by the scaling factor
Crop input data		Sets the limitation of the data used for the calculation to only the cropped selection defined above. If this option is not selected, the entire image set is used for the calculation.

Additional Parameters

Additional settings for Input ROI expand, when you click in the operation.

Parameter	Description
Bounds	Sets the analysis area edges. The whole XY bounds, the viewing area or a custom space can be applied.
Planes	Sets the analysis planes range. A single plane, a range of planes or the whole stack can be selected.
Time Points	Sets the analysis time points range. A single time pint, a range of time points or the whole movie can be selected.

Machine Learning Segmenter

Automatic objects detection algorithm based on machine learning approach. The ML training (Nuclei and Membrane) was previously created.

In most cases, the default settings are suitable, and you don't need to change them. You can modify a few special parameters, related to the segments, by clicking the Operation menu > Segments.

Machine Learning Segmenter parameters

Parameter		Description
Training		Shows the name of the Training that is loaded and some basic information below that.
Channels		Sets the channels as input to this filter. The software will select the channels automatically based on the channels used for training.
Outputs		Configure the outputs for this segmenter such as the output and classes name or the coloring of found segments.

Object Feature Filter

The results of the Machine Learning Segmenter are filtered by volume to delete artifacts.

Object Feature Filter parameters

Parameter		Description
Input		Sets the filter input. If more than one segment operation is present in your pipeline, the correct input source must be set.
Is of type		Sets the type of filter which filters the objects based on the selected type. By default, it is set on Any for all object types. The list of available features is updated accordingly.
Filter selection		Sets the function for the segmentation is performed.
Criterion Settings		These parameters depend on the selection of the filter.
+ Add Filter		Allows to add a Simple, Ratio or Tag filter criterions.

Store Objects

Store the detected segments (tag) in the active dataset.

Executing the pipeline

You can execute the pipeline step by step or in a single run. To do this, use the executing buttons in the Pipeline toolbar.

Executing step by step

You can execute the pipeline step by step (back and forth). This method allows to run and undo a single operation. You can either use the executing buttons in the Pipeline toolbar or in the Operation toolbar to go through the operation list.

To run the single operation, click .
To undo the single operation, click .
Note: Undo the last operation executed if you need to change the operation settings.

Executing in a single run

As alternative to executing step by step, execute the pipeline in a single run.

To run the whole pipeline, click .
To stop the pipeline execution, click .

Operation status

Executing buttons in the Operation toolbar after executing

When the operation is running, this icon is shown.

When the operation is completed, this icon is shown.

Modifying the current pipeline

You can modify the pipeline to adapt to another datasets. Therefore, all the pipeline parameters should be set according to the new dataset features.

Adjusting the operations

The parameters of each operation are described here: Pipeline operations layout

You have opened the pipeline.

Go to the operation you want to change.
Set the appropriate parameters as desired.
Execute the pipeline (see Executing the pipeline).

Adding or removing operations

You can add or remove operations from the sample pipeline.

Click + Add operation...
The operation list is grouped in four groups by their typology.
To add the operation to the current pipeline, double-click it. Alternatively, drag & drop the operation to the current Pipeline.
Note: The operation cannot be added during the Pipeline execution.
The operation will be inserted at the end of the group of operations to which it belongs.
Voxel operations are positioned before the segment generation. Store operations are always put at the end of the Pipeline.
To remove an operation, click Close at the Operation toolbar.

Carl-Zeiss-Strasse 22

73447 Oberkochen

Germany

Detecting nuclei and cells (Machine Learning)