CheXagent
Stanford AIMI Chest X-ray Foundation Model
Powered by Dr7.ai Medical AI Platform
CheXagent is an 8-billion-parameter vision-language model from the Stanford Center for Artificial Intelligence in Medicine & Imaging (AIMI) built specifically to interpret chest X-rays â from finding detection and report generation to visual question answering.
Try CheXagent-Style Chest X-ray Analysis
Upload a chest radiograph and ask about findings, views or follow-up â the demo shows the kind of structured chest X-ray interpretation CheXagent was designed for.
Try Other Models
View AllCheXagent Demo
Experience Advanced Medical AI
Unlock Full CheXagent Potential
Get unlimited chest X-ray analysis, higher upload limits and API access for radiology research workflows.
What is CheXagent?
CheXagent is a chest X-ray foundation model released in January 2024 by Stanford AIMI (Chen et al., "CheXagent: Towards a Foundation Model for Chest X-Ray Interpretation"). It combines a clinical language model, a vision encoder trained on chest radiographs, and a bridging network that lets the model reason about images in natural language.
To train it, the team built CheXinstruct â a large instruction-tuning dataset assembled from dozens of public chest X-ray sources â and CheXbench, an evaluation suite that scores models on image perception tasks (view classification, disease identification, visual QA) and on text generation tasks (findings and impression generation, report summarisation).
On CheXbench, CheXagent outperformed general-purpose and earlier medical multimodal models on most tasks, and the authors also audited it for performance differences across sex, race and age. It remains a research model: it is not FDA-cleared and is not intended to replace a radiologist.
Latest Development
Released January 2024 by Stanford AIMI (Chen et al., 2024)
Weights and the CheXbench evaluation suite are available for research; not a cleared medical device
Key Features
Built and benchmarked specifically for chest radiographs
Core Capabilities
Chest X-ray view and orientation classification
Disease and finding identification
Findings and impression (report) generation
Visual question answering on chest X-rays
Radiology report summarisation
Trained on CheXinstruct, a large multi-source instruction dataset
Evaluated on CheXbench across perception and generation tasks
Fairness audit across sex, race and age subgroups
Performance Benchmarks
CheXbench results are reported in the CheXagent paper; Dr7.ai has not independently measured them
CheXbench â perception
LeadingView classification, disease identification and VQA versus general and medical multimodal baselines
CheXbench â generation
LeadingFindings, impression and summarisation tasks scored with radiology-specific metrics
CheXinstruct
Multi-sourceInstruction dataset built from many public chest X-ray collections (e.g. MIMIC-CXR, CheXpert)
Model size
8BVision encoder + vision-language bridger + clinical LLM
Use Cases
Imaging Research
Benchmark chest X-ray interpretation models with a common suite (CheXbench).
Radiology Decision Support
Draft findings and flag likely abnormalities for radiologist review.
Radiology Education
Interactive question answering on teaching chest X-rays.
How to Use CheXagent
Research weights from Stanford AIMI
Get the Model
CheXagent weights are published by Stanford AIMI on Hugging Face for research use; check the model card for the current licence terms.
Integration Options
Feed a chest radiograph plus an instruction and receive text.
- âĸ Finding / disease identification prompts
- âĸ Findings and impression generation
- âĸ Visual question answering
- âĸ Report summarisation
Deployment Considerations
Plan for privacy, validation and human oversight before putting CheXagent anywhere near patient data.
- âĸ De-identify data and comply with HIPAA / GDPR
- âĸ Validate on your own institution's data before use
- âĸ Keep a licensed clinician in the loop for every output
- âĸ Document limitations and intended use for auditors
Important Considerations
Clinical Validation Required
CheXagent is a research model. Its outputs must be reviewed by qualified healthcare professionals before any clinical use.
Regulatory Compliance
The model is not a cleared or approved medical device. Ensure compliance with local healthcare regulations and data-protection rules (e.g. HIPAA, GDPR) before deployment.
Have a photo of a visible symptom?
The models described on this page are research tools for clinical imaging. If what you have is an everyday photo â a rash, a mole, a nail, a wound â AI Doctor Scan reads it and returns a free urgency level and a plain-language summary in about 30 seconds. It does not name a diagnosis, and it needs no account.
Try AI Doctor Scan â freeCheXagent vs Other Medical Imaging Models
A specialist versus generalists
CheXagent
Stanford AIMI's chest X-ray model
- âPurpose-built for chest radiographs
- âHandles perception and report generation in one model
- âPublic benchmark (CheXbench) for fair comparison
- âFairness evaluation across demographic groups
- âTrained on a large multi-source instruction set
- âOpen for research use
Chest X-ray research, report-drafting prototypes, and benchmarking radiology AI
Other Medical AI Models
How it compares
- ÃGeneral LLMs (GPT-4, Gemini) are stronger at open-ended reasoning but are closed and costly
- ÃLarger open models (Meditron-70B, MedGemma-27B) need far more GPU memory
- ÃText-only models cannot look at images
- ÃImage-only classifiers cannot explain their findings in language
- ÃCommercial products are validated but not inspectable
- ÃMost alternatives publish fewer reproducible benchmarks
Pick by modality (text vs. image), licence, and the hardware you actually have
Frequently Asked Questions
Common questions about CheXagent
Who developed CheXagent?
The Stanford Center for Artificial Intelligence in Medicine & Imaging (AIMI). It is not a Google model.
What can CheXagent do?
View classification, finding identification, visual question answering, and generating or summarising findings and impressions for chest X-rays.
Is CheXagent FDA-cleared?
No. It is a research model and must not be used as the sole basis for a clinical decision.
What is CheXbench?
The evaluation suite released with CheXagent: eight chest X-ray tasks spanning image perception and text generation, used to compare multimodal models.
Does this demo run CheXagent itself?
The demo uses Dr7.ai's multimodal medical assistant to show chest X-ray interpretation. For the exact model, use the Stanford AIMI weights.