GCP-Net: A Gating Context-Aware Pooling Network for Cervical Cell Nuclei Segmentation

Accurate segmentation of cervical nuclei is an essential step in the early diagnosis cancer. Still, there are few studies on clustered clusters cells. Because complexities high cell overlap, blurred boundaries, and cells, accurate remains a pressing challenge. In this paper, we purposefully propose GCP-Net deep learning network to handle challenging cluster images. The proposed U-Net-based consists pretrained ResNet-34 model as encoder, Gating Context-aware Pooling (GCP) module, modified decoder. GCP module primary building block improve quality feature learning. It allows refine details maps leveraging multiscale context gating Global Context Attention for spatial texture dependencies. decoder including Attention- (GCA-) Residual Block helps build long-range dependencies global interaction predicted masks. We conducted extensive comparative experiments with seven existing models our ClusteredCell dataset three typical medical image datasets, respectively. experimental results showed that obtained promising evaluation metrics AJI, Dice, PQ, demonstrating superiorities generalizability automatic comparison some SOAT baselines.