Endo-MedSAM: Pelvic MRI Dataset for Uterus Segmentation in Endometriosis

A pelvic MRI dataset of 74 subjects and 3,449 T2-weighted slices from two institutions for developing AI models for uterus segmentation in endometriosis. The dataset was used to fine-tune the Endo-MedSAM model, achieving mean 3D Dice scores of 0.81–0.88 with bounding-box prompts. The dataset was uploaded by Rawan AlSaad on figshare in May 2026.

Use Cases

• Fine-tuning vision foundation models for medical image segmentation based on MRI slices.
• Evaluating AI model performance across institutions and prompting strategies based on multicenter and single-center cohorts.
• Generating standardized volumetric measurements for endometriosis care based on 3D reconstructions.
• Benchmarking zero-shot versus fine-tuned model performance for uterus delineation based on Dice and HD95 metrics.

Strengths

• Contains 3,449 T2-weighted MRI slices from 74 subjects, providing a substantial image base.
• Includes data from two institutions (multicenter and single-center), allowing for cross-institutional evaluation.
• Model performance is documented with specific metrics, achieving mean 3D Dice scores of 0.81–0.88 with bounding-box prompts.

Limitations

• Row count and column-level documentation are absent; field semantics must be inferred after download.
• The dataset is 188.7 KB, which suggests limited scope and likely contains metadata or a supplementary PDF rather than raw image files.
• Freshness should be verified; the last update date is 2026-05-25, which is a future date.

Provenance

Source

figshare

Collection Method

MRI scans collected from two institutions (D1, multicenter; D2, single-center).

Freshness

Last updated 2026-05-25 12:50:56; freshness should be verified.