Project Summary: Annotated cells in pathology images, developed algorithms to expand point annotations into region-based annotations, and trained/tested benchmark models to evaluate performance

● Annotated cells in large-scale pathology images, ensuring accurate identification of different cell types for downstream analysis

● Designed and implemented algorithms to transform point-based annotations into region-based masks, enabling richer training data for segmentation tasks

● Trained and evaluated benchmark models including CellViT, Hover-Net, and StarDist on the curated dataset, systematically analyzing performance across multiple metrics

• Developed FishLEN, a deep learning model for estimating fish length from images to support biological research.

• Built an end-to-end pipeline utilizing data preprocessing, model training, and evaluation with PyTorch.

• Enhanced model performance through iterative experimentation, including data augmentation and model tuning.

Project Summary: Reviewed research papers in the field, analyzed experimental results from models based on provided GitHub files, and presented findings through visualizations

● Reviewed and analyzed papers to summarize cutting-edge models in cell classification and segmentation,

deepening understanding and tracking advancements

● Ran the Hover-Net project hosted on GitHub from a published paper, thoroughly analyzed the output results, and extracted key features to facilitate deeper insights

● Visualized the extracted feature data comprehensively, generated images, and converted selected features into QuPath-compatible files to support further analysis and applications