ACS-SegNet: An Attention-Based CNN-SegFormer Segmentation Network for Tissue Segmentation in Histopathology

  • 2025-10-23 17:21:06
  • Nima Torbati, Anastasia Meshcheryakova, Ramona Woitek, Diana Mechtcheriakova, Amirreza Mahbod
  • 0

Abstract

Automated histopathological image analysis plays a vital role incomputer-aided diagnosis of various diseases. Among developed algorithms, deeplearning-based approaches have demonstrated excellent performance in multipletasks, including semantic tissue segmentation in histological images. In thisstudy, we propose a novel approach based on attention-driven feature fusion ofconvolutional neural networks (CNNs) and vision transformers (ViTs) within aunified dual-encoder model to improve semantic segmentation performance.Evaluation on two publicly available datasets showed that our model achieved{\mu}IoU/{\mu}Dice scores of 76.79%/86.87% on the GCPS dataset and64.93%/76.60% on the PUMA dataset, outperforming state-of-the-art and baselinebenchmarks. The implementation of our method is publicly available in a GitHubrepository: https://github.com/NimaTorbati/ACS-SegNet

 

Quick Read (beta)

loading the full paper ...