Inverse problems and image reconstruction in acoustics and elastography
Contact person: Sven Peter N?sholm
Keywords: Image reconstruction, inverse problems, wave propagation, elastograpy, neural operators
Research group: Digital Signal Processing and Image Analysis (DSB)
Department of Informatics
The theme involves the use of acoustic time-series and array data for the characterization of a range of media and sources. This includes the general aspect of solving ill-posted inverse problems using classical, full-waveform inversion, and machine-learning based approaches, as well as adaptive beamforming algorithms. In the medical domain, the key thematic applications is in medical ultrasound image reconstruction, as well as ultrasound-based elastography. In the underwater domain, the key thematic application is active sonar and in particular synthetic aperture sonar. In the atmospheric acoustic domain, the key application is in using infrasound waves to probe the upper atmospheric circulation. We also appreciate contributions related to using operator-based partial differential equation numerical schemes for machine-learning supported speedup of linear and nonlinear wave propagation simulations. For the elastorgaphy path, we seek to estimate myocardial stiffness using, for example, ultrasound time harmonic elastography with methods inspired from MR elastography approaches.
Methodological research topics:
- Full waveform inversion approaches to estimate tissue properties using medical ultrasound datasets - also using wave-propagation speedup using neural operator approaches
- Improvement of shear wave velocity estimation by using different ultrasound scanning schemes
- Comparison of time harmonic elastography with methods utilizing naturally induced shear waves in the heart in 2-D and 3-D
- Underwater active sonar for 1) carbon capture and storage, where monitoring the integrity of subsea CO2 storage is sought after using acoustics based gas bubble detection; 2) synethetic aperture sonar for high-resolution seabed imaging
Application domains:
- For enhanced ultrasound image reconstruction and tissue property inversion: diagnostic ultrasound across several medical domains
- For ultrasound based elastography: Cardiovascular diseases are the leading cause of death globally and a noninvasive, robust, and cost-effective diagnosis method has great impact potential as a clinical tool that can be additional to estimating blood velocities in the heart to assess the overall cardiac function.
- For infrasound-based atmospheric probing: a better representation of the upper atmospheric winds in numerical weather prediction and atmospheric models has great potential for enhancing medium-range weather prediction.
External partners:
- Oslo University Hospital (OUH)
- Norwegian Computing Center (NR)
- Kongsberg Maritime AS