PACS & Imaging
SASTRO PACS
Web-based PACS for small hospitals and primary clinics.
DICOM 3.0MODALITY WORKLISTWEB VIEWER
Productionv5.4
AI & ResearchResearchv0.9
MedicVue AI Lab
Research and evaluation of AI models for chest imaging.
A research workspace for public CXR datasets, preprocessing, evaluation, and tuberculosis detection model development.
CXR
Dataset
ONNX
Export
TB
Research focus
MedicVue AI Lab - OverviewSASTRO
AI OVERLAY · RESEARCHCONFIDENCE
TB92%
Edema12%
Cardio6%
NOT A CLINICAL DIAGNOSIS
Overview
Research and evaluation of AI models for chest imaging.
MedicVue AI Lab is a research workspace for processing public chest X-ray datasets, building consistent preprocessing pipelines, and evaluating models. Its initial focus is tuberculosis detection, later used in DICOM Station.
Models that pass evaluation are exported to ONNX so they can run on-device. The lab is not a clinical product: its output is a model candidate that must be validated further before any medical decision.
Key value
- Public CXR dataset pipeline and preprocessing.
- Reproducible model evaluation metrics.
- ONNX export for on-device inference.
- Tuberculosis detection research focus.
Features
Designed for real workloads.
Each feature group answers one part of the facility workflow.
01
Data
- Public CXR dataset curation
- Preprocessing and normalization
- Training, validation, and test splits
- Dataset source and license documentation
02
Model
- Training and fine-tuning
- Metric evaluation and confusion matrix
- ONNX export
- Model and configuration versioning
03
Validation
- Hold-out data testing
- False positive and false negative analysis
- Documented model limitations
- Research disclaimer, not a diagnosis
Specifications
Technical specifications & system requirements.
A summary of the stack and deployment modes. Full details are available in a technical session.
| Language | Python |
|---|---|
| Runtime | ONNX Runtime |
| Dataset | Public CXR |
| Output | ONNX model |
| Status | Research |
| Use | Not a clinical product |
Integrations
Connects to the ecosystem you already use.
ONNX RuntimeDICOM StationCXR Dataset
Frequently asked
Implementation & hospital IT questions.
No. The model is at research stage and must not be used as a diagnostic basis without clinical validation.
More products
Other parts of the ecosystem.
Clinical Infrastructure Consultation
Ready to map your facility's needs?
A 30-minute session to map the radiology, hospital system, and monitoring workflow you need.