KUB-StoneX: 2,570 Kidney-Ureter-Bladder X-rays with Stone Segmentations

KUB-StoneX is a publicly available, de-identified clinical imaging dataset containing 2,570 anteroposterior kidney-ureter-bladder (KUB) radiographs from 1,703 patients. It was retrospectively collected from two tertiary care hospitals in Bangladesh by Sabiba Israt Zerin and includes expert-verified pixel-level stone segmentation masks, region-of-interest patches, and pre-extracted radiomic features. The dataset was last updated on 2026-04-30.

Use Cases

• Developing deep learning models for urinary stone detection based on annotated KUB X-ray images.
• Benchmarking segmentation algorithms using the provided expert-verified pixel-level stone masks.
• Training and evaluating radiomics models for stone characterization based on the pre-extracted radiomic features.
• Researching hybrid AI models that combine imaging and radiomic data for low-cost X-ray-based diagnosis.

Strengths

• Contains 2,570 annotated KUB radiographs from 1,703 patients.
• Includes expert-verified pixel-level segmentation masks and pre-extracted radiomic features.
• Data is organized in a standardized directory structure with consistent file naming for direct correspondence.
• The dataset is 19.3 GB in size, indicating a substantial collection of imaging data.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect geographic bias inherent to its collection from two hospitals in Bangladesh.
• Row count is unknown, which may limit suitability assessment for certain modeling tasks.

Provenance

Source

Retrospectively collected from the Kidney Foundation Hospital and Research Institute, Mirpur-2, and the National Institute of Kidney Diseases and Urology (NIKDU), Bangladesh.

Collection Method

Images were fully anonymized. Stone regions were annotated by experienced radiologists using polygon-based labeling, with consensus masks generated from multiple sessions.

Freshness

Last updated 2026-04-30 18:42:40; freshness should be verified.

Geography

Bangladesh