Third Opinion AI

Chest X-ray
Artificial intelligence for radiologist workflow improvement
We create the World where every radiologist has access to innovative tools for quick and qualitative patient examination
Identified conditions for diagnosing
Semiotic features classification developed by leading russian radiologists allows to use Third Opinion AI algorithms to diagnose a wide range of pathological changes in chest X-ray
Flu and Pneumania:
Infiltration (AUC ROC 87%)
Dark patch center (AUC ROC 93%)
Pulmonary pathology (AUC ROC 89%)
Hydrothorax (AUC ROC 98%)
Pneumothorax (Dev)
Atelectatic Changes (Dev)
Tumors:
Dark patch center (AUC ROC 93%)
Bone tumor(Dev)
Respiratory T.B.:
Dark patch center (AUC ROC 93%)
Calcinate / Calc. shadow (AUC ROC 94%)
Hydrothorax (AUC ROC 98%)
Dissemination (Dev)
Cavity with Decay and Liquid Level (Dev)
Pneumothorax (Dev)
Atelectatic Changes (Dev)

Bone pathology:
Fracture (AUC ROC 87%)
Spine Pathology(AUC ROC 90%)
Development Anomalies (Dev)
Bone tumor(Dev)
Value
Doctor Time-saving
Radiologist work optimization will reduce time used for a test in average by 40%
Certificate accuracy
Today every third radiologist conclusion requires a review. Third Opinion will bring diagnostic accuracy closer to 100%
Timely assistance
The service implements prioritization of radiologist's worklist,allowing timely assistance for patients with severe diseases
How it works
Studies circulation

Service analyzes in background images and prioritizes worklist
Third Opinion
Algorithm creates automatically additional DICOM series in study indicating found feature and its localization
DICOM Structured Report
Research results are saved in DICOM SR format for quick access from PACS-Viewer
Pre-filled conclusion
Service forms a pre-filled conclusion based on study results to reduce radiologist routine work
Counterparts