CVC‐EndoSceneStill - sporedata/researchdesigneR GitHub Wiki

General description

The CVC-EndoSceneStill database is a publicly available dataset that has been specifically developed to aid in research related to gastrointestinal endoscopy, particularly the study and development of computer vision algorithms for colorectal cancer detection. This dataset has been produced by the Computer Vision Center (CVC) in Spain and is primarily used to advance the field of medical imaging by training and evaluating AI and machine learning models to perform various tasks such as polyp detection, segmentation, and classification.

Limitations

1. Data Imbalance: Similar to other medical imaging datasets, there can be an imbalance in the number of polyp versus non-polyp images, which can affect the model’s ability to learn effectively.

2. Complexity of Annotations: Manually annotating polyps, especially the segmentation masks, is a time-consuming process that requires experts. The quality of these annotations is crucial for the development of accurate AI models.

Related publications

• Towards automatic polyp detection with a polyp appearance model
• A benchmark for endoluminal scene segmentation of colonoscopy images
• WM-DOVA maps for accurate polyp highlighting in colonoscopy: validation vs. saliency maps from physicians