Automated Nucleus Segmentation in Histology Images: An Attention U-Net Approach with Comprehensive Preprocessing and Inference Pipelines
Objectives: In the area of digital pathology, nucleus segmentation is an essential step for quantitative analysis; however, high nuclear density and staining heterogeneity in histology images pose a great challenge. Methods: This study implements a lightweight Attention U-Net model on the MoNuSeg 2018 dataset to overco...