NBIPolyp UCdb - sporedata/researchdesigneR GitHub Wiki

General description

The NBIPolyp-UCDB dataset is a specialized medical image dataset focused on the detection and classification of polyps in the gastrointestinal (GI) tract, particularly for aiding in the diagnosis of colorectal cancer. This dataset is essential for developing and validating computer-aided diagnostic (CAD) systems, particularly those using deep learning and machine learning techniques, to improve the accuracy and efficiency of polyp detection during colonoscopy.

The dataset contains 86 endoscopic images captured using Narrow-Band-Imaging (NBI). These images focus on colorectal polyps, which are critical in the early detection and prevention of colorectal cancer.

The NBIPolyp-UCDB dataset includes a wide variety of polyp types, shapes, and sizes, ensuring a comprehensive dataset for training AI models. The dataset usually contains labeled images, specifying whether a polyp is present or absent and may provide further classification into types of polyps.

Limitations

  1. Data imbalance: There might be fewer images with polyps compared to those without, which can introduce challenges in training AI models.
  2. Lighting and visibility: Factors like lighting, the angle of the endoscope, and obstructions in the colon (e.g., folds or fecal matter) can complicate image analysis.
  3. Variability in appearance: Polyps can vary significantly in size, shape, and texture, making detection a challenging task.

Related publications

Data access

For more information on the NBIPolyp-Ucdb dataset, visit https://www.mat.uc.pt/%7Eisabelf/Polyp-UCdb/NBIPolyp-UCdb.html