Deep learning model for analyzing tissue displacement from ultrasound imaging
This technology is a deep learning approach for estimating tissue displacement in the heart and arteries from ultrasound images that can be utilized to improve the performance of elasticity imaging techniques.
Unmet Need: Robust quantitative analysis of tissue displacement
Current techniques for measuring tissue displacement are unable to handle the complex motion of tissue and lack quantitative analysis. Tissue displacement measurements are therefore under-utilized in clinical applications such as diagnosing and monitoring cardiovascular diseases and certain forms of cancer.
The Technology: Commercially compatible deep learning technique to quantify tissue displacement in real time
This technology is a deep learning model called Voxelmorph that is trained on ultrasound displacement images of the heart and arteries to improve the performance of elasticity imaging techniques. The model learns the physiological displacement patterns of cardiovascular tissues to improve pulse wave imaging and myocardial elastography performance. Ultrasound images used for training are acquired using commercially available instruments and analyzed in MATLAB. This approach is compatible with commercial ultrasound machines that use elasticity imaging modalities, enabling real time analysis of ultrasound images to generate tissue displacement maps.
This technology has been validated using human common carotid arteries in vivo.
Applications:
- Diagnostic imaging for cancer and cardio-vascular diseases
- Clinical tool for real-time measurements of tissue displacement
- Clinical tool for improved performance of Pulse Wave Imaging and Myocardial Elastography
- Research tool for characterizing cardiovascular diseases
Advantages:
- Compatible with commercial ultrasound systems
- Captures complex tissue motion patterns
- Provides robust and quantitative tissue displacement measurements
- Enables real-time tissue displacement analysis
Lead Inventor:
Related Publications:
Tech Ventures Reference:
IR CU21356
Licensing Contact: Dovina Qu