U-Net live : Segmentation network for F-actin nanostructures in STED images of living neurons
Authors/Creators
- 1. Université Laval
Description
Segmentation network for F-actin nanostructures (rings and fibers) in STED images of living neurons. The model was trained on STED images of fixed neurons that were adapted to match the features of live-cell STED images with TA-CycleGAN.
Input : STED image (20 nm/px)
Output : Two-channels segmentation prediction for F-actin rings (first channel) and fibers (second channel)
networks.py : Functions to generate the model, adapted from https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix
UNet_live_model.py : Script to build the model, load the weights, and test on an example image
pretrained_UNet_live.pth : Trained network weights
20201216_Auto_Select_MultiRegioncs4_ROI2_initial.tif : Example STED image to test the model
Publication : Resolution Enhancement with a Task-Assisted GAN to Guide Optical Nanoscopy Image Analysis and Acquisition (https://doi.org/10.1101/2021.07.19.452964)