
PaxeraHealth’s Enterprise Solutions with AI technology at their core are revolutionizing the process of anomaly identification, optimizing workflows, automating tasks, streamlining access to patient data at meaningful points of care, and improving care coordination.
Increase reading accuracy by 7% with PaxeraAI
Human-Centered Design
Paxera’s AI model is designed to learn and improve its accuracy as it interacts with radiologists, using the latest machine-learning technology to monitor users’ behavioral patterns and adjust to their preferences, in addition to contributing their diagnostic knowledge to the model’s database.
With thousands of built-in commands, EraBot is able to interact seamlessly with users via speech and text, gives them access to radiology resources during reading or dictation, as well as pulls relevant patient data from the EMR directly on the viewing screen alongside images, simplifying access to lab results, medications, patient history, and more.
Faster Triage of Cases
PaxeraAI provides real-time CDS (clinical decision support) with augmented reading aids, localization visualization of identified lesions via region outlines or heat maps, as well as abnormality scores. Acute abnormalities are detected and marked as they enter the work list, allowing radiologists to focus on prioritizing life-threatening cases.

81%
Accuracy
Mammo AI detects lesion locations, malignant and benign classifications (architectural distortions, calcifications, masses), density, and BI-RAD categories. Trained with 13,360 high quality annotated images.
80%
Accuracy
CT AI detects Intracranial Hemorrhage and its subtypes, including: Epidural, Subdural, Intraparenchymal, Intraventricular, and Subarachnoid. Trained with 36,000 high quality annotated images.
87%
Accuracy
Chest AI detects 10 classes including consolidation, pleural effusion, atelectasis, nodule, mediastinal widening, pneumothorax, and more. Supports Tuberculosis screening. Trained with 600,000 high quality labelled and annotated images.
Training & Validation
PaxeraAI is trained on hundreds of thousands of studies, using the collective knowledge of the medical community to inform future analysis. As with common evaluation techniques for radiologist performance, PaxeraAI models are evaluated primarily in terms of AUC (area under the ROC curve) for classification of malignancy, as well as precision-recall (PRAUC) and mean average precision (mAP). In addition to labeled data and annotated studies, the lesion detection system also integrates DICOM tags and patient history to inform its algorithms.


Reduce reading time
and fast track findings

Increase confidence
in results

Boost reporting and
outcome efficiency