GAA-DETR-breast: Mammography Images with Radiologist Gaze and Lesion Annotations

Breast Lesion Detection Gaze Dataset is a subset of a large-scale medical dataset presented at MICCAI 2025. It combines high-resolution mammography images with expert radiologist eye-tracking data and gold-standard lesion annotations. The dataset was created by yankon0408 and was last updated on March 26, 2026.

Use Cases

• Training AI models for lesion detection based on high-resolution mammography images.
• Studying radiologist visual search patterns based on expert eye-tracking data.
• Developing gaze-assisted or gaze-aligned computer vision models based on the combined image and gaze data.

Strengths

• Dataset is associated with a MICCAI 2025 Spotlight paper, indicating peer-reviewed scientific relevance.
• Combines three data modalities: high-resolution images, expert gaze data, and lesion annotations.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

yankon0408 on Hugging Face

Collection Method

Likely collected from clinical settings with expert annotations and eye-tracking equipment.

Freshness

Last updated 2026-03-26 07:58:13; freshness should be verified.