Real-time AI System for Echocardiography Analysis and Quantification

Summary:

We are seeking statements of capability or interest from parties interested in collaborative research to further develop, evaluate or commercialize this technology.

Description of Technology:

Scientists have developed an artificial intelligence (AI) system to automatically extract predictive biomarkers from echocardiography. This AI system takes an echocardiography study as input and produces predictive biomarkers as output in real-time. It performs several automated image analysis tasks including echocardiography quality (quality for the acquired echo clips) assessment, echocardiography view retrieval (e.g., 2D four chamber or parasternal long axis view, Doppler), echocardiography cardiac region segmentation (e.g. inferior vena cava) and cardiac biomarker quantification (e.g. right atrial pressure). This system is:

1. efficient (in terms of time and space)
2. robust (can handle missing data)
3. interpretable (can provide human-understandable insights)

Such features allow its use in embedded devices for real-world clinical practice and point of care testing. This technology can be used to assist the diagnosis of a wide range of common cardiac diseases as well as to evaluate cardiac function in minority populations such as sickle cell disease patients. 

Potential Commercial Applications:

• Cardiologists in clinical practice, to:
  • streamline cross-functional workflows
  • obtain temporal biomarkers of a full study in a single run
• Cardiology researchers, to:
  • Enable systematic comparison across millions of echocardiograms
  • rapidly analyze cardiac data across hospitals and research centers
  • perform analysis and quantification for pattern detection, verification, and informed hypothesis testing
• Perform diagnosis and assessment of wide range of cardiac diseases including coronary artery disease, heart failure, and cardiopulmonary complications in sickle cell disease patients
• Once converted into a software application, can be installed in bedside and handheld ultrasound devices to extract imaging biomarkers
• Can combine the extracted imaging biomarkers with non-imaging biomarkers (e.g., lab tests, age)
• Applicable to other clinical fields; e.g., endocrinology

Competitive Advantages:

• Real-time medical imaging
  • No comparable system in the market that performs a series of medical image tasks in real-time to extract predictive imaging biomarkers from different echo types and views including Doppler, M-model, and B-mode (A2C, A4C, PLAX, IVC, etc.)
• Enhance efficiency and save time – hours to minutes or seconds
• Enable deeper analysis of a patient’s condition
• Enables the streamlining of tedious clinical tasks (e.g., image analysis tasks and visual biomarkers observation), potentially saving hours of physicians’ time
• Reduced costs to payers and patients resulting from decreased physician time
• Reproducible and precise image analysis anywhere
• Enables unprecedented portability and standardized access to care — including visual analytics, predictive biomarker calculation, diagnosis, assessment, and prognosis