Aivia Software

(Aivia 12.1) 3D Neuron Analysis - FL

Owned by Hung-yu Chang
Last updated: Sept 30, 2024 by Won Yung Choi
9 min read

The 3D Neuron Analysis - FL recipe in Aivia detects and automatically traces dendrites in fluorescence 3D volumetric images. The recipe has options for tracing dendrites with or without somas as well as options for automatic detection of dendritic spines. After applying the recipe, you can extend the analysis by applying a neuron classifier to classify the detected neurons.

For tracing neurons in 3D electron microscopy images, use the 3D Neuron Analysis - EM recipe. 

After Aivia 13, new measurements are added for the neuron level:

Intensity measurements for neuron
-Mean, Max, Min for neuron (all components)
-Mean, Max, Min for Neuron (dendrites)
-Mean, Max, Min for Neuron (spines)

After Aivia 12.1, users can

  1. Use Cellpose (1) model as the soma detector to detect neurons in dense regions.

  2. Choose different input channels for soma detection and dendrite tracing.

  3. Filter out elongated false positive objects

Automatic Parameter Selection

Auto Icon

3D Neuron Analysis - FL features Aivia That Learns: AI-powered automatic parameter selection.

Click on the Auto icon (see right) in the recipe header to initiate parameter prediction, which feeds features from the specified input image channel into a model trained by human experts and then populates the Recipe Console with predicted parameters.

Auto icon for automatic parameter selection

 

Active Tile

Another way to initiate parameter estimation for 3D Neuron Analysis - FL is through the Neuron Analysis Active Tile, which is accessed from the search bar at the top of the Aivia window. The Neuron Analysis Active Tile also adjusts the Aivia layout to one that is recommended for a neuron analysis recipe workflow. If the Neuron Analysis Active Tile is not shown when you click in the search bar, try clicking on the Expand icon in the lower-left corner of the search dropdown.

Active Tiles

Parameters and Presets

Input and Output

After Aivia 12.1, the recipe can take two different input channels for soma and dendrite/spine detection.

Parameters

Recipe parameters for 3D Neuron Analysis - FL and their descriptions are summarized in the table below.

Preset Group

Parameter Name

Min Value

Max Value

Description

Preset Group

Parameter Name

Min Value

Max Value

Description

Soma Detection

Soma Diameter

0

1,000 (px or µm)

Specifies the average diameter of somas (neuron cell bodies); a lower value leads to detection of smaller objects on the image

Max Soma Aspect Ratio

0

100

Soma will be removed from output set if the ratio of the major axis to the average of two minor axes exceed the threshold

Switchable Option: Detect Soma with Cellpose

Using Cellpose based deep learning model as soma detector. Suitable for images with a high density of neurons.

Seed Detection Intensity Threshold

0

255 (8-bit)

65,535 (16-bit)

This parameters is used to set the minimum soma intensity as a threshold to detect seeds for preview and skip block processing. *For Cellpose for soma detection option, the preview option generates spots for indicating locations where Cellpose will be run for improved speed for preview options but these spots do not 100% correspond to where somas get generated as Cellpose may not detect somas at the location where the preview spots are generated.

Cell Probability Threshold

0

100

Only probability values above the threshold in the Cellpose predicted probability map will under go further processing to generate soma mask.

Switchable Option: Detect Soma with Partition

Blob detection with 3D watershed

Min Soma Intensity Threshold

0

255 (8-bit)

65,535 (16-bit)

Specifies the minimum intensity of somas on the image for partitioning touching somas; this parameter is only enabled when the Enable Soma Partition option is selected; a lower value reduces the number of partitions and preserves larger soma

Switchable Option: Skip Soma Partition

Skip the blob detection and using Otsu thresholding as soma detector

Dendrite Detection

 

Dendrite Diameter

0

50 (px or µm)

Specifies the diameter of a typical dendrite on the image; a lower value increases detection sensitivity for thin dendrites and decreases detection sensitivity for thicker dendrites (such as axons)

Min Branch Length

0

1,000 (px or µm)

Specifies the minimum length of a traced branch to be included in the analysis results

Intensity Threshold

0

255 (8-bit)

65,535 (16-bit)

Specifies the intensity range for dendrite tracing using the voxel scooping algorithm; the lower threshold specifies the minimum intensity that is not considered background; the upper threshold specifies the minimum intensity above which the voxels are always considered as part of a dendrite

Mean Intensity Offset

0

255 (8-bit)

65,535 (16-bit)

Adjusts the detection sensitivity of the dendrite tracing algorithm; a higher value reduces the number of dendrites detected that are at or near the minimum intensity threshold specified above

Switchable Option: Skip Soma Detection

 

Max Connection Distance

 

0

1,000 (px or µm)

Specifies the maximum gap distance between dendritic traces to join together; this parameter is only enabled when the Skip Soma Detection option is selected; a lower value increases the number of disconnected dendritic branches

Spine Detection

(Enable Spine Detection and Enable Enhanced Spine Detection only)

Typical Spine Head Diameter

0

100 (px or µm)

Specifies the approximate diameter of a typical dendritic spine head; this parameter is only enabled when the Enable Spine Detection option or Enable Enhanced Spine Detection option is selected

Spine Head Diameter

0

100 (px or µm)

Specifies the size range for a spine to be included in the analysis results based on the diameter of the spine head; this parameter is only enabled when the Enable Spine Detection option or Enable Enhanced Spine Detection option is selected

Max Spine Neck Length

0

100 (px or µm)

Specifies the maximum distance between spine heads and dendritic branches; this parameter is only enabled when the Enable Spine Detection option or Enable Enhanced Spine Detection option is selected; a lower value reduces the search range for spine heads

 

Presets

There are three preset groups in the recipe: Soma Detection, Dendrite Detection, and Spine Detection; each group has three pre-configured parameter groupings to help you get started on the analysis. The default preset values are given in the sections to follow.

Soma Detection

Parameter Name

Small

Medium

Large

Parameter Name

Small

Medium

Large

Soma Diameter

10 - 50 (px or µm)

40 - 100 (px or µm)

100 - 300 (px or µm)

Min Intensity Threshold

61 (8-bit)

15,728 (16-bit)

41 (8-bit)

10,486 (16-bit)

10 (8-bit)

2,621 (16-bit)

 

Dendrite Detection

Parameter Name

Small

Medium

Large

Parameter Name

Small

Medium

Large

Dendrite Diameter

2 (px or µm)

5 (px or µm)

15 (px or µm)

Min Branch Length

10 (px or µm)

20 (px or µm)

50 (px or µm)

Intensity Threshold

10 - 50 (8-bit)

2,621 - 12,845 (16-bit)

40 - 80 (8-bit)

10,289 - 20,578 (16-bit)

60 - 100 (8-bit)

15,401 - 25,690 (16-bit)

Mean Intensity Offset

0

0

0

Max Connection Distance

10 (px or µm)

25 (px or µm)

50 (px or µm)

 

Spine Detection

Parameter Name

Small

Medium

Large

Parameter Name

Small

Medium

Large

Typical Spine Head Diameter

1 (px or µm)

2 (px or µm)

5 (px or µm)

Spine Head Diameter

0.05 - 4 (px or µm)

0.2 - 10 (px or µm)

1 - 15 (px or µm)

Max Spine Neck Length

2 (px or µm)

2 (px or µm)

2 (px or µm)

Spine Detection is disabled by default. Expand the preset group and select Enable Spine Detection or Enable Enhanced Spine Detection to enable the preset group and view the parameters. With Enable Enhanced Spine Detection, spines are detected with greater accuracy at the expense of processing time.

 

 

Tutorial

Before beginning the tutorial, please download the Neuron Analysis Demo image. For information on how to select presets or modify parameter values, please refer to the tutorial on how to use the Recipe Console.

  1. Unzip the demo file and load the demo image, "NeuronAnalysisDemo.tif," into Aivia.

  2. In the Recipe Console, click on the Recipe selection dropdown menu and select the 3D Neuron Analysis - FL recipe.

  3. Click the drop-down arrow icon beside Input and Output

    1. Select the Unnamed channel in the demo image as the Input Soma Channel

    2. Select the Unnamed channel as the Input Dendrite Channel as well

  4. Click on the Show Advanced Interface icon to expand the Recipe Console and show the parameter options for the recipe.

  5. Modify the parameter values in the recipe as follows for the Soma Detection and Dendrite Detection groups:

    • Soma Diameter: 8 - 15

    • Max Soma Aspect Ratio: 100

    • Select Skip Soma Partition

    • Dendrite Diameter: 1.5

    • Min Branch Length: 5

    • Intensity Threshold: 18 - 200

    • Mean Intensity Offset: 0

  6. To enable spine detection, go to Step 6; otherwise, go to Step 8.

  7. Click on the Switch Recipe Operations icon for the Spine Detection group to show the list of available recipe operations; select the Enable Spine Detection option from the dropdown menu.

  8. Select the Small preset for the Spine Detection group.

  9. Click the Start button or press the F4 key on your keyboard to begin applying the recipe to the image.

The recipe generates three output object sets: Soma Set, Dendrite Set and Spine Set. You can toggle the display of individual object sets as well as change their display colors from the Object Set Settings panel.

 

Results

 

 







Measurements

The 3D Neuron Analysis - FL recipe generates morphological, intensity, and count measurements for the somas, dendrites, and spines of each detected neuron. You can add additional measurements to the analysis results by using the Measurement Tool in Aivia and view measurement definitions on the Measurement Definitions page. Select Measurements generated by the recipe are described in the table below.

Object Set

Morphology

Intensity

Count

Object Set

Morphology

Intensity

Count

Soma Set

  • Surface Area

  • Volume

  • Mean Intensity

  • Max Intensity

  • Min Intensity

  • Total Intensity

  • Std. Dev. Intensity

None

Dendrite Trees

  • Mean Diameter

  • Total Path Length

  • Longest Path Length

  • Shortest Path Length

None

  • Branch Count

  • Spine Count

  • Branch Order

Dendrite Segments

  • Mean Diameter

  • Branch Angle

  • Total Path Length

  • Longest Path Length

  • Shortest Path Length

  • Surface Area

  • Volume

  • Average Length

  • Top 5 Percent Length

  • CV Lengths

  • Average Angle

  • Mean Intensity

  • Max Intensity

  • Min Intensity

  • Total Intensity

  • Std. Dev. Intensity

  • Branch Count

  • Spine Count

  • Branch Order

Spines

  • Total Spine Length

  • Head Neck Length Ratio

  • Total Spine Volume

  • Total Spine Surface Area

  • Head Neck Volume Ratio

  • Spine Head Diameter

  • Spine Diameter Ratio

  • Length to Head Diameter Ratio

  • Spine Surface Area Ratio

  • Spine Convex Hull Volume Ratio

  • Spine Tortuosity

  • Branch Angle

None

None

Spine Heads

  • Surface Area

  • Volume

  • Average Length

  • Top 5 Percent Length

  • CV Lengths

  • Average Angle

  • Mean Intensity

  • Max Intensity

  • Min Intensity

  • Total Intensity

  • Std. Dev. Intensity

None

Spine Necks

  • Surface Area

  • Volume

  • Average Length

  • Top 5 Percent Length

  • CV Lengths

  • Average Angle

  • Mean Intensity

  • Max Intensity

  • Min Intensity

  • Total Intensity

  • Std. Dev. Intensity

None

Neurons

  • Min Max, Mean, Total Dendrite Branch Mean Diameter

  • Min, Max, Mean, Total Dendrite Branch Path Length

  • Min, Max, Mean, Total Dendrite Branch Surface Area

  • Min, Max, Mean, Total Dendrite Branch Volume

  • Min, Max, Mean, Total Dendrite Branch Tortuosity

  • Min Max, Mean, Total Dendrite Branch Local Bifurcation Angle

  • Min, Max, Mean, Total Dendrite Branch Partition Asymmetry

  • Min, Max, Mean, Total Dendrite Branch Path Length to Root

  • Mean, Max, Min for neuron (all components)

  • Mean, Max, Min for Neuron (dendrites)

  • Mean, Max, Min for Neuron (spines)

  • Dendrite Count

  • Nunber of Bifurcation

  • Number of Branches

  • Number of Terminal Tips

  • Min, Max, Mean, Total Dendrite Branch Node Count

  • Min, Max, Mean, Toital Dendrite Branch Generation

 

Image Credits

Maryann Martone and Eric Bushong, Cell Image Library (CIL:48401), http://www.cellimagelibrary.org/images/48401


Related Articles

References

  1. Stringer C, Wang T, Michaelos M, and Pachitariu M. Cellpose: a generalist algorithm for cellular segmentation. Nature Methods. 18: 100-106. (2021)