28 Nov 2022

AI Assisted Tool for CryoEM Segmentation

Here we provide a tutorial to interactively segment a cellular cryo-electron tomography image using the AI assisted segmentation tool of the Segmentation+ workroom.

The AI Assisted Segmentation tool helps users to produce segmentations on complex images with minimal effort. This interactive approach uses a shallow neural network to provide results as fast as possible, with the AI Assisted Segmentation tool doing most of the work segmenting the image. It is particularly well suited to speeding up the processing of electron microscopy images but can also be used with other 3D datasets.

In this tutorial, the AI Assisted Segmentation tool is used to interactively segment and reproduce the results of the EMD-8594 cryo-EM image that was provided in [1].  First, a minimal part of the image is labelled, and patches added that will be used to train the AI-assisted segmentation tool. After the first result, the user can gradually improve the segmentation using the AI Assisted segmentation again or the other tools available in the Segmentation+ workroom.

References:

[1] Muyuan Chen, Wei Dai, Stella Y Sun, Darius Jonasch, Cynthia Y He, Michael F Schmid, Wah Chiu & Steven J Ludtke, Convolutional neural networks for automated annotation of cellular cryo-electron tomograms, Nature Methods 14; 983-985, 2017.