Figure 4
From: Deep learning method for comet segmentation and comet assay image analysis
Overview of the proposed framework for comet analysis. In the training stage, a deep neural network is trained using a comet image and a corresponding mask image to enable the model to precisely segment each comet in the image. Besides, it enables the classification of the ghost and non-ghost cells. In the testing stage, the trained comet segmentation model is first used to segment and classify each comet. Then, followed by applying a series of post processing techniques, a general comet scoring method is applied to score them and place the results together to generate a result table for each comet image.
