AI enhanced image analysis
Boost your analysis with AI
We are bringing a whole new level of flexibility and power to your current image analysis pipeline by allowing you to embed deep learning into the process.
​
In Aivia 9.5, each Aivia recipe now includes an optional deep learning pre-processing step. The result of this step is then directly passed along to the recipe for further analysis, yielding a seamless way to enhance and analyze your images.
​
-
Enhance any Aivia recipe with a deep learning (DL) pre-processing step
-
Apply DL models as a recipe
-
Easily batch apply DL models
-
Online library of pre-trained DL models
Additionally, deep learning models can be applied to any image via the recipe console. Open or drag-and-drop the model into the recipe console and it automatically converts to a recipe for you. This provides you an easy way to deploy and batch apply any pre-trained deep learning model from our library or created in Aivia.
​
Using this new system you can add the power of AI smoothly into your existing analysis workflow.
Media Gallery
Image credit: Chen et al, 2020
AI image restoration
and virtual staining
Leverage our expertise
Over the last 3 years we have worked closely with top researchers to develop impactful AI models for the microscopy community. Now we are releasing these ready-to-use models to the wider community.
​
We are providing 20 new pre-trained deep learning models covering 7 different organelles and 3 imaging modalities. The application of these models fall into multiple categories: image restoration, super resolution, deconvolution, image segmentation and virtual staining. Most of these models are validated and characterize in our recent pre-print, check it out here.
-
20 new pre-trained DL models ready to use
-
Covering 7 organelles and 3 imaging modalities (iSIM, confocal, STED)
-
Applications include image restoration / deconvolution, image segmentation and virtual staining
-
Model library to explore and search for the right model for you
These models are available to you (the Aivia user) and can be accessed and used in multiple ways. If your data looks similar to our training data you can deploy it in Aivia as a recipe, as a pre-processing step in an application specific recipe (e.g. 3D Cell Analysis), or in Aivia Cloud. If your data is different, you can modify our model slightly (transfer learning in Aivia Cloud) or use our model structure to train a brand-new model (regular Aivia Cloud training of DL model).
​
You can access the pre-trained models from within Aivia (Help>Update>Models). We also organized these models (including a description and test image data) into a online library so you can explore and find the best model for your application. Just search the library for your organelle, imaging modality or application.
​
With this large selection of pre-trained deep learning models, you don’t have to be an expert to use AI.
​
​
Media Gallery
Image credit: Chen et al, 2020