520,041 unique CXR scans were used from across a diverse patient and care setting. Each scan was independently read 3 times from a group of 148 qualified Radiologists. 124 findings were established as present or not present across each scan, leading to a total of 198,135,621 CXR labels.

Our internal evidence demonstrates that Annalise Enterprise CXR is non-inferior to radiologists in detecting 124 different findings, and when used by radiologists, improves their performance in 96 findings. This evidence was published in peer-reviewed study in The Lancet Digital Health.

Currently Annalise CXR is only available for use in chest x-rays and covers 124 findings. AI products for additional modalities are planned for future releases.

Current radiological findings of COVID are diverse and non-specific. Annalise CXR can detect findings relating to COVID pneumonia including airspace opacity and interstitial changes. Although we don’t claim to diagnose COVID pneumonia on CXR alone as we believe that it’s still too inaccurate to be clinically useful.

Please refer to our privacy policy for more information.

The partnership with I-MED has shared approximately 400 million de-identified medical images and 30 million de-identified reports used by radiologists to detect and diagnose conditions.

Our AI systems have been developed for radiology. Annalise detects radiological findings. Diagnosing disease is the responsibility of physician. See the complete list of 124 chest x-ray findings here.

Annalise CXR interfaces with picture archiving & communication systems (PACS) and radiology information systems (RIS) to obtain the chest X-ray images to process. The Artificial Intelligence (AI) algorithm within the device uses deep learning techniques to identify the presence of the radiological findings. Additionally, Annalise CXR analyses the chest radiographs using deep learning techniques to identify the laterality or highlight the relevant areas of interest for a subset of findings as defined in the user guide. The suspected findings are communicated to the clinician viewing the study by displaying the findings and associated localisation information to the clinician as they view the study in the PACS viewer.

Deep learning is part of a broader family of machine learning methods based on artificial neural networks. In deep learning, the neural networks have various deep layers that enable learning. From vast quantities of data, deep learning algorithms enable machines to solve complex problems even if a data set is diverse, unstructured and interconnected. The more data that deep learning algorithms have to learn from, the better they perform.

Machine learning is a type of artificial intelligence. It involves programming a computer so it can make accurate predictions based on an available dataset. Examples of machine learning include image recognition, traffic predictions and suggested search engine results.

Artificial intelligence (AI) is branch of computer science that attempts to build machines capable of intelligent behaviour.