FAST: Filamentous Actin Segmentation Tool for quantifying cytoskeletal organization

We developed Filamentous Actin Segmentation Tool (FAST), that leverages deep learning and antibody assisted labeling to segment and quantify actin structures from optical microscopy images. This is the model that was trained with the dataset Dataset Description (see Version V1): `train_data` contains the images used for training the model where `images` subdirectory contains the images used in training the model, `predicted_masks` are predictions from the model post training, and `ground_truth_masks` contains corresponding ground truth masks. `test_data`contains the images used for evaluating the model where `images` sub-directory contains the images that are not used in training the model, `predicted_masks` are predictions from the model post training, and `ground_truth_masks` contains corresponding ground truth masks. `alternate_cells/llcpk1` contains the images from LLC-PK1 cells used for evaluating the model where `images` sub-directory contain the images, `predicted_masks` are predictions from the model post training, and `ground_truth_masks` contains corresponding ground truth masks. `alternate_cells/3t3` contains the images from NIH-3T3 cells used for evaluating the model where `images` sub-directory contain the images, `predicted_masks` are predictions from the model post training, and `ground_truth_masks` contains corresponding ground truth masks. `lifeact_3t3` contains the images from NIH-3T3 fibroblasts expressing LifeAct-GFP used for evaluating the model where `control_dmso_images` subdirectory contain the images from vehicle control DMSO, `control_dmso_predicted_masks` are model predictions for corresponding images, `rock_inhibitor_images` subdirectory contain the images from cells treated with ROCK inhibitor Y-27632, and `rock_inhibitor_predicted_masks` are model predictions for corresponding images.