Milvue detection detects seven commonly found pathologies in Emergency Department Chest and MSK X-rays namely pneumothorax, lung nodule, opacity, pleural effusion, fracture, dislocation, and elbow fat pad. The model pushes worklist flags to RIS/PACS and EMR and draws bounding boxes around the suspected region of illness.
To assist emergency physicians and radiologists in detecting the following pathologies in Emergency Department Chest and MSK X-ray: pneumothorax, lung nodule, opacity, pleural effusion, fracture, dislocation, and elbow fat pad.
Milvue is trained to detect and classify only the seven pathologies listed in the Description section. Diseases outside of the seven are not recognized by the model.
Milvue is not trained with intensive care unit or operating room X-rays. Utilizing the model in these settings is outside the scope of the intended use.
Information on training data
Trained with over 30,000 professionally annotated X-ray data, taken by eight OEM machines, including GE, Siemens, and Phillips, at over 5 different institutions in France.
Model performance metrics
Area ROC (AUC) : 0.87 - 0.98 (depending on the pathology)
Inference Time: 1 second
Model available on Arterys.
Milvue, the # 1 RN solution in radiology that detects abnormalities, prioritizes exams and alerts the doctor.