Dataset Overview
Definition：Mammography X-Ray Imaging Datasetis provided byChangsha Langhui Information Technology Co., Ltd.A Large-Scale X-Ray Imaging Database Built. Based on 300,000 Cases. Final settlement is based on the de-publiced count of accidental exports/Studies (diagnostic administrative reports are still accounted with the same example). Accomplishable X-Ray (Mammography) check as the subject, using the DICOM standard Raw Data Format, to maintain complete sequence information, spatial geometry parameters and Metadata. Overing more than 10 cooperation cases, from multi-center sources, with strict Quality control; capable for mediamaging AI model training and assessed-diagnosis capacity building.
| Dataset Name | Mammography X-Ray Imaging Dataset |
| Total Data Volume | 300,000 Cases (counted toward the total X-ray volume of 200,000 cases)。Final settlement is based on the de-duplicated count of accepted bilateral examinations / Studies (diagnostic unilateral supplementary views are still counted within the same examination)。(Bilateral Examination / Study Level (diagnostic unilateral supplementary views are still counted within the same examination)) |
| Imaging Modality | X-Ray (Mammography) |
| Scan Protocol | Double-sided CC; double-sided MLO; diagnostic complementary view; digital mammography/DBT (if any); synthesis of 2D maps separated from original projection. The protocol name must be consistent with the actual document, scope, period, view or sequence. |
| Raw Data Format | De-identified raw DICOM with full hierarchy and metadata preserved |
| Core Conditions (P0) | Breast masses, suspicious calcifications, focal / global asymmetry, architectural distortion, breast cancer, etc. |
| Extended Conditions (P1) | Screening negative |
| Source Institution | It is recommended to cover 10–20 institutions, with a minimum of 10 in principle and no upper limit；The specific number of partner Grade-A tertiary hospitals is adjusted flexibly according to the delivered data volume |
| Quality Control Standards | Kappa ≥ 0.75, five-tier L1–L5 quality grading |
| Use Cases | AI model training, assisted diagnosis, radiomics research, algorithm validation |
Disease Distribution Overview
Delivery Specifications and Field Requirements
The dataset strictly follows unified delivery index requirements to ensure data quality and traceability。
Delivery & Counting：One row corresponds to one bilateral examination / Study (diagnostic unilateral supplementary views are still counted within the same examination)；Multiple sequences, phases, reconstructions, views or repeated exports within the same exam may not be counted separately。
Composition of Positive Cases：Abnormal / positive examinations as the main body；Screening-negative cases must be established as an independent control group and stratified by BI-RADS and follow-up status, with the proportion fixed in the purchase order。Findings such as "suspected", "considered", "cannot be excluded" and "post-treatment changes" must be retained separately from pathologically confirmed diagnoses and must not be forcibly merged into a definite diagnosis。
Coverage：Standard bilateral CC and MLO views as the main body；Supplementary compression / magnification views, implant views and tomosynthesis must be labeled in separate layers。Positive, negative or disease labels are generated only for structures that are actually within the diagnostic field of view and whose image quality supports evaluation。
Raw Data Requirements：de-identified original DICOM or equipment is a non-destructive digital image, with exposure parameters, pixel spacing, position/view, plant model and original/processed state; film flipping, report amplification and social software compression are not allowed to replace. Amputation, film photographs, report PDF or key frame puzzles are not allowed to replace the agreed original image/video/slice.
Field Group Coverage：Batch source and location, anonymized patient information, exam primary key, modality and body part, protocol and device, specialized fields, file hierarchy, report text, diagnostic labels, clinical context, related fields, quality and disposition
Structured Text：At minimum, retain the exam name, exam findings, diagnostic conclusion/impression, primary diagnosis, fine-grained labels, and negative and uncertain semantics；When original reports are available, they must be linked case by case。
Data Volume Growth Trend
2026 Frontier AI Research Progress
Domain Review：The X-ray series of AIs from 2025 to 2026 entered the basic model age, and large-scale pre-training models made breakthroughs in multi-mission broad-based and small sample learning. The top journals Nature, Nature Medicine, The Lancet and others published several clinical-level validation studies that facilitated the transition of the X-ray series of AIs from laboratory to clinical deployment.
DeepMind Mammography AI Clinical Validation
Nature, 2025DeepMind Breast AI achieved a 19 per cent reduction in false positives and a 20 per cent reduction in pseudo negatives in over a million screening cases, which was consistent in the validation of multiple European multi-country centres.
Mammography Screening AI Large-Scale Clinical Deployment
The Lancet Oncology, 2026The national breast screening study in Sweden showed that AAA-assisted screening had increased cancer detection rates by 13 per cent and that the rate of inter-stage cancer had decreased significantly.
Multi-View Mammography Transformer
IEEE TMI, 2025Transformer structure based on a double view integration of CC/MLO, with radiologists in BI-RADS classification and breast cancer risk prediction.
Digital Breast Tomosynthesis AI
Radiology, 2026DBT 3D breast-tectonic synthesis AI detection system, which is 25% higher than traditional 2D molybdenum detection in the meth-memplified population.
Annotation Workflow & Quality Control
Image Acquisition & De-identification
Standardized acquisition workflow: complete data is exported directly from the devices, and patient identifiers (PHI) are removed before storage to ensure data compliance。
Initial Annotation (Specialist Physician)
Attending physicians annotate case by case against the standard, including lesion localization, morphological description, disease diagnostic labels and specialized indicators。
Review (Associate Chief Physician or Above)
Experts with the title of Associate Chief Physician or above review each initial annotation result item by item, correcting erroneous annotations and supplementing missing dimensions to ensure annotation accuracy。
Consistency Assessment
10% of samples are randomly selected and independently annotated by 3 physicians to calculate Fleiss' Kappa; below 0.Dimensions scored 75 are flagged for rework。
Quality Acceptance Checklist
License and Usage Agreement
Academic Research License
For universities and research institutions, supporting academic research related to medical AI。After signing the agreement, a de-identified data subset is provided; the source must be acknowledged：Langhui Technology DataAssetsAPI。
Commercial License
For medical device companies and AI diagnostics companies, supporting medical AI product R&D and medical device registration。Provides full datasets + custom annotation + incremental update services。
Data Compliance Statement
All images come from legally authorized sources; PHI fields are fully de-identified and contain no information that can directly identify an individual, in compliance with the Personal Information Protection Law and the Data Security Law。
Customized services
Supports extended requirements such as adding specific disease types, multimodal annotation expansion, and custom training of AI-assisted diagnostic models。
Quick Facts
AI Frontier Research
The X-ray series 2025-2026 AI entered the basic model phase, with multi-modular self-monitoring learning performing excellently in a number of benchmark missions, and several top-level journal papers promoted clinical AI diagnostic development.