Lizard: A Large-Scale Dataset for Colonic Nuclear Instance Segmentation and Classification

• DOI: 10.48550/arxiv.2108.11195
• Publication Date: 2021-01-01
• Authors: Simon Graham; Mostafa Jahanifar; Ayesha Azam; Mohammed Nimir; Yee‐Wah Tsang; Katherine Dodd; Emily Hero; Harvir Sahota; Atisha Tank; Ksenija Benes; Noorul Wahab; Fayyaz Minhas; Shan E Ahmed Raza; Hesham El Daly; Kishore Gopalakri...; David Snead; Nasir Rajpoot
• Keywords: Ground truth

The development of deep segmentation models for computational pathology (CPath) can help foster the investigation interpretable morphological biomarkers. Yet, there is a major bottleneck in success such approaches because supervised learning require an abundance accurately labelled data. This issue exacerbated field CPath generation detailed annotations usually demands input pathologist to be able distinguish between different tissue constructs and nuclei. Manually labelling nuclei may not feasible approach collecting large-scale annotated datasets, especially when single image region contain thousands cells. However, solely relying on automatic will limit accuracy reliability ground truth. Therefore, overcome above challenges, we propose multi-stage annotation pipeline enable collection datasets histology analysis, with pathologist-in-the-loop refinement steps. Using this pipeline, generate largest known nuclear instance classification dataset, containing nearly half million H&E stained colon tissue. We have released dataset encourage research community utilise it drive forward downstream cell-based CPath.