URI-CAD: Ultrasound Renal Images with Kidney Annotations, 1985 Images

URI-CAD is a 2D ultrasound imaging database containing 1985 B-mode renal images (450 healthy, 1535 pathological). The dataset was retrospectively collected from patients aged 18 and above at Hospital Universitario Ramón y Cajal between 2009 and 2018. Annotations by two experienced nephrologists include polygonal kidney segmentation masks, indicators of global pathologies, and bounding boxes for local lesions.

Use Cases

• Training segmentation models to locate kidneys in ultrasound images based on the provided polygonal masks.
• Developing classifiers to detect global kidney pathologies using the annotated diagnosis labels.
• Building object detection models to identify and localize specific lesions from the provided bounding boxes.
• Benchmarking CAD systems for renal ultrasound using a dataset with expert annotations and a balanced distribution of healthy and pathological cases.

Strengths

• 1985 annotated images with a clear breakdown of 450 healthy and 1535 pathological kidneys.
• Annotations provided by two experienced nephrologists, suggesting clinical expertise.
• Balanced distribution across patient sex and left/right kidneys, and images were anonymized for privacy.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The temporal coverage from 2009 to 2018 may not reflect current imaging practices or pathologies.

Provenance

Source

Hospital Universitario Ramón y Cajal, IRYCIS

Collection Method

Retrospectively collected B-mode renal ultrasound images, one image per clinical case, anonymized.

Time Range

2009 to 2018

Freshness

Last updated 2025-10-14 21:26:48; freshness should be verified.

Geography

Data collection location is likely Spain, based on the hospital name, but not explicitly stated.