Fig. 5: Performance comparison of UNSEG, Cellpose, and Mesmer for the entire GIT dataset.
From: UNSEG: unsupervised segmentation of cells and their nuclei in complex tissue samples
First row compares median F1 score performance curves for the three methods as a function of IoU threshold for nucleus and cell segmentation of images in the GIT dataset. The insert contains median F1 score values at the IoU threshold of 0.5 for three algorithms. The second and third rows, respectively, show pairwise comparison between UNSEG and Cellpose, and UNSEG and Mesmer. The comparison includes median F1 score curves along with their 95% confidence intervals. Their complete overlap indicates similar performance of all three methods.
