Nuclei detection is a key task in Ki67 proliferation index estimation in breast cancer images. Deep learning algorithms have shown strong potential in nuclei detection tasks. However, they face challenges when applied to pathology images with dense medium and overlapping nuclei since fine details are often diluted or completely lost by early maxpooling layers. This paper introduces an optimized UV-Net architecture, specifically developed to recover nuclear details with high-resolution through feature preservation for Ki67 proliferation index computation. UV-Net achieves an average F1-score of 0.83 on held-out test patch data, while other architectures obtain 0.74-0.79. On tissue microarrays (unseen) test data obtained from multiple centers, UV-Net's accuracy exceeds other architectures by a wide margin, including 9-42\% on Ontario Veterinary College, 7-35\% on Protein Atlas and 0.3-3\% on University Health Network.
@article{arxiv.2111.05482,
title = {Preserving Dense Features for Ki67 Nuclei Detection},
author = {Seyed Hossein Mirjahanmardi and Melanie Dawe and Anthony Fyles and Wei Shi and Fei-Fei Liu and Susan Done and April Khademi},
journal= {arXiv preprint arXiv:2111.05482},
year = {2022}
}
Comments
Published in SPIE Medical Imaging 04/2022: Digital and Computational Pathology; 120390Y. Event: SPIE Medical Imaging, 2022, San Diego, California, United States