Power to Discover
Complete Neuron Reconstruction
In Aivia 5.1, we are extending our novel neuron detection approach so that it successfully analyses more types of samples. With the newest version of the 3D Neuron Analysis recipe, you can trace dendritic networks that lack a visible soma (because it is not labelled or is out of the field of view) and you can automatically detect dendritic spines. Regardless of the types of neuron analysis you need, you will experience the same fast automatic tracing performance you have come to expect in Aivia.
The Neuron Editor has also been significantly improved and expanded. The new version of the Dendrite Pencil automatically centers your traces and determines the radius of the traced segments in real time. With a single click, you have the option to manually detect cell bodies (somata in the case of neurons) and spines; Aivia will automatically determine their size and shape. With Aivia 5.1, it takes less time than ever to reconstruct whole neurons from fluorescent 3D images.
See below an example of a neuron reconstruction done completely in Aivia.
Count Cells and Nuclei in 3D
From 5.1 you can use Aivia to automatically detect fluorescently labeled 3D objects. Using the latest image segmentation technology Aivia uses intensity, size and shape input to detect and count objects in 3D. Its advanced algorithms work well in both sparsely populated samples and in data sets where objects (e.g. cells or nuclei) are touching each other. Aivia can automatically discern the boundary between two adjacent objects by carefully looking for tell tale object-object transition features (e.g. sudden drop and rise in the intensity of an object). Segregating and exploring the characteristics of sub populations of objects is possible by applying single parameter (e.g. cell volume) sorting/filtering and binning. In addition, Aivia includes an object type/phenotype classification tool powered by state of the art machine learning. The machine learning workflow is designed to be used by life science and imaging experts, requiring no background in mathematics, machine vision or machine learning. The cell classifier make use of all measured parameters (unlike the single parameter sorting/filtering mentioned above) and is thus much more capable of distinguishing complex and subtle phenotypes.
Machine Learning Phenotyping
Image segmentation and object detection results in the generation of a plethora of intensity, morphology and texture related measurements. In the past, researchers were satisfied using a small subset of parameters known to dominate a particular phenotype. However, our machine learning approach harnesses the full wealth of information contained in each and every measurement. This is the technology behind our fast and robust automatic object classification.
The machine learning-enabled classifier allows you to train the classification algorithm to look for specific phenotypes (e.g. developmental stages, disease states and cell types.) present in your data. In addition to neuron classification, Aivia 5.1 supports classification of individual dendrite segments, dendrite branches, spines, and other 3D objects such as cells and nuclei.
In Aivia 5.1, it is possible to import SWC, VRML, and IV object reconstructions from other analysis applications into Aivia's machine learning classifier. This way, you can create new classifiers or augment an excisting one using reconstructions / analyses done previously.
Because our approach uses all the data available and is automated it has the potential to minimize human-introduced biases and increase productivity.
For more information, watch our Insight webinar on machine learning approaches to cell discovery below.
Major Under-the-Hood Improvements
In Aivia 5.1, we have made significant improvements for the rendering and viewing of millions of 3D objects. The first thing you will discover upon launching Aivia 5.1 is the introduction of the dark-mode GUI. The dark GUI provides better viewing experience of fluorescent image data sets, especially in dark rooms where the imaging happens.
Next, you will discover significant speed boost when rendering your data sets with millions of 3D objects already created. You can navigate across the entire 3D scene with minimal frame rate loss. The speed boost allows scenes in virtual reality (VR) to be rendered more smoothly - reducing the risk of motion sickness while you explore the analysis results in the immersive VR environment.
Check out this virtual reality fly-through of a neuron detected and viewed in Aivia 5.1.