Overview

Overview

Biodock gives any research team the ability to create and use an AI model that analyzes complex microscopy images and extracts the metrics needed to accelerate their research, even without an extensive programming background.
We allow scientists to create ground truth datasets and train powerful custom AI in one platform without writing a single line of code.
Once complete, the model is accessible through the Analysis dashboard as a one click AI analysis pipeline.
​Analysis tab with different AI models that have been trained

Creating an AI pipeline

Creating an AI algorithm requires two key steps, image annotation and training. Once complete, the model is accessible through the Analysis dashboard as a one click AI analysis pipeline.
In Biodock, AI pipelines are built within projects. Projects organize a team and a group of image data together to collaboratively train AI models. They also help manage versioning and status of these models. We recommend creating one project per type of image.

1. Image annotation

Supervised AI can be defined as an AI model that uses labels associated with ground truth annotations to classify or predict regions of interest. A supervising AI algorithm is very effective at segmenting images to detect boundaries of cells and tissue regions. It is, therefore, necessary to generate a labeled dataset of microscopy images containing accurate and representative ground truth labels. Please refer to Best Practices page to review how to create accurate and representative labels.
There are 3 steps to creating a pipeline - creating a project, labeling your images, and training. Find the three sections on how to do this below: