/
Neuron Analysis - Golgi-Stained Neurons

Aivia Software

Neuron Analysis - Golgi-Stained Neurons

 

Background: Golgi staining is a classical staining technique, first published in 1873, and widely used in the field of neuroscience for morphological studies of neurons due to its unique ability to stain a small number of neurons (estimated to be approximately ~1-3% of total neurons1) in their entirety, making it possible for visualizing the entire structure of neurons. While widely used in neuroscience for decades, the method still relies on brightfield imaging which creates two specific challenges for downstream image analysis: 1) objects that out of focus are captured as blurry objects on multiple planes and 2) as Golgi stain is dark and opaque, physical shadow from neuronal somas and dendrites is cast on optical planes further reducing the signal to noise ratio.

The steps outlined below show a method to overcome the challenges mentioned above to trace Golgi-stained neurons:

1) how to use our patented machine-learning Pixel Classifier to generate an enhanced image channel that removes blurred noises around the dendrites,

2) to apply a 3D neuron analysis workflow to automatically detect somas using the Otsu method and trace the dendrites using the voxel scooping method.

Tutorial

  1. Launch Aivia Launchpad and select the Golgi-Stained Neuron button to launch the Guided Sequence for Golgi-stained neuron tracing.

     

  2. Open an image with Golgi-stained neurons and select the Enhancement step.

image-20250423-215313.png
A 2D single z-plane view of Golgi-stained neurons

  1. Invert image to create an “Inverted” channel.

image-20250423-215758.png

 

 

 

  1. Train a Pixel Classifier using a small kernel size and line features. Paint on dendrites (link for in-depth tutorial on the Pixel Classifier), ensuring to paint out-of-focus dendrites as background and selecting planes where the dendrites are in sharp focus. The somas can be left as background. Choose the Inverted channel as the input:

image-20250423-220402.png
Pixel Classifier Kernel Size and Feature Selection to “Small” and “Line”.
image-20250423-221219.png
Out of focus dendrites trained as background
image-20250423-221320.png
In-focus dendrites painted as “Want”

 

image-20250423-221451.png
“Want” channel created with the Pixel Classifier with dendrites in focus

  1. Run the 3D Neuron Analysis Recipe with the “Inverted” channel as the soma channel input and the newly created “Want” channel for the dendrites:

image-20250424-021001.png

 

  1. Traced neurons. Inspect and manually correct as needed using the Neuron Composer (link):

image-20250424-021411.png
Golgi-stained neurons traced for morphological characterization and downstream classification

 

  1. Explore the morphological characteristics of the dendrites, somas, and spines (optional to detect) using the spreadsheets, charts, and classifiers.

 

•1Zaqout S, Kaindl AM. Golgi-Cox Staining Step by Step. Front Neuroanat. 2016 Mar 31;10:38. doi: 10.3389/fnana.2016.00038. PMID: 27065817; PMCID: PMC4814522.