EVICAN

EVICAN provides a curated dataset for training and evaluating deep learning models for cell and nucleus segmentation from microscopy images.

Key Features:

• Diverse Image Collection: 4,600 grayscale images from 30 different cell lines captured with various microscopes, contrast mechanisms, and magnifications.
• Annotated Segmentations: Approximately 26,000 segmented cells annotated for segmentation tasks.
• Brightfield Imaging Coverage: Images include brightfield microscopy suitable for automated cell and nucleus segmentation.

Scientific Applications:

• Deep learning training and evaluation: Use as training and validation data for cell and nucleus segmentation algorithms, including Mask R-CNN.
• Quantitative image analysis: Develop automated segmentation workflows for studies in cancer research, developmental biology, and drug discovery.

Methodology:

The dataset was used with a Mask R-CNN implementation, achieving a mean average precision (mAP) of 61.6% with a Jaccard Index above 0.5.