AI-Powered Neuron Analysis
Improve capacity to automatically trace neurons in challenging conditions
Aivia 12 enables deep learning-based soma detection for densely populated neuronal images based on the generalist Cellpose algorithm (Stringer, 2021) [1] that has been adapted for neuronal soma detection.
Additionally, machine-learning based neuron analysis makes it possible to leverage on the user’s expertise on the neuronal morphology to train the software and to automatically trace Golgi-stained neurons for downstream morphological characterization including dendritic branching analysis via 3D Sholl analysis, and 3D spatial analysis between neurons and other structures of interest.
-
Cellpose for better soma detection in densely populated neuron images
-
Pixel classifier + 3D neuron analysis recipe for detection in Golgi-stained images
-
Automated neuron tracing with separate input for soma and dendrites
-
Preview for soma detection
[1] Stringer C, Wang T, Michaelos M, Pachitariu M. Cellpose: a generalist algorithm for cellular segmentation. Nature Methods. 18: 100-106. (2021)
Media Gallery
Relational Measurement
Relational measurement between structures of interest
With the implementation of Object Relation Tool, structures of interest, regardless of the object type, can be explored for distance and density measurements in relation to each other.
-
3D distance measurement for the nearest object from another object (e.g. dendrite to vesicles)
-
3D density measurement for the nearest objects
-
Count of all objects within a user-specified distance from another object
Media Gallery
Accessible Deep Learning Models
Accessible AI for Discoveries
AI-Based Parameter Estimation with Aivia That Learns (ATL) has been iteratively improved for better prediction for challenging data. Our model now generalizes to a greater variety of images to yield an even more accurate parameter set for 3D object analysis.
Additionally, 4 pre-trained deep learning models from BioImage.io are ready to be used via simple drag and drop into Aivia so that users can begin to use deep learning models without needing to train their own models, set up their own Python and CUDA environments, or learn to code.
With Aivia, anyone can get started with AI without expertise in deep learning.
-
Improved generalizability of ATL parameter prediction
-
Drag and drop apply of 4 deep learning models from Bioimage.io within Aivia to leverage on all the existing user-friendly workflows and visualization tools
Media Gallery
Insight with Data Exploration
Better insights via data exploration
The latest version of Aivia also includes additional Improved data exploration capabilities, including new charting tools that allow for more comprehensive data visualization and analysis.
-
Violin plot/Pearson correlation coefficient
-
Polygon gating for new group creation
