Meeting Banner
Abstract #2435

Noise Robust Inverse Laplacian Operator Based Reconstruction of Global Elastic Parameters in Magnetic Resonance Elastography.

Sebastian Papazoglou1, 2, Heiko Tzschaetzsch1, Juergen Braun3, Ingolf Sack1

1Radiology, Charit University Medicine, Berlin, Germany; 2NeuroCure Clinical Research Center, Charit University Medicine, Berlin, Germany; 3Medical Informatics, Charite - University Medicine Berlin, Berlin, Germany

Magnetic resonance elastography (MRE) provides diagnostically significant average viscoelastic parameters in organs with good signal to noise ratio (SNR) such as, e.g. the human liver. Usually, parameter reconstruction in MRE requires the computation of noise amplifying derivatives and also involves the inversion of noisy wave data. Therefore this direct approach is ineffective in case of very low SNR, e.g. in MRE on the human lung. In this study we present an approach based on the wave equation featuring the inverse Laplace operator, which requires no inversion to robustly recover global viscoelastic parameters even in presence of strong noise.

Keywords

added affected allows although approaches appropriate assess assessment avoids background bias biasing boundary circumvent clinical complex component conditions conjugate consequence considering consistent constant constitution contained contrast coordinates correct denominator denotes density dependence derivatives detection develop direct displacement distinct domain drive elastic elasticity employed entire equation equations field fixed formulate formulation frequency fully generic global gradient harmonic held hepatic imaginary in vivo incorporates increasing influence informatics inherently instable introduced inverse inversion investigation investigations invokes known least lengths linear little mass matrix medicine medium modality modulus much noise noisy normally omitting operator oscillating overdetermined overestimates part particularly performance pixel pixels radiology reads real reason reasonably reconstructed reconstruction recovered regularly related renders required resolution respect retained rewritten sample scalar sensitivity shear similarly simulated simulations soft solve spatially square squares strong strongly studied suffers suggesting system tissue transpose true underestimated underestimates underestimation useful varying vector vibration view viscoelastic wave wavelength wavelengths waves ways whether whole width yields