PURPOSE
Investigate the potential of magnetic resonance elastography (MRE) to differentiate and delineate tissues with different viscoelastic properties, with the purpose of applying MRE to differentiate and delineate recurrent brain tumor and radiation necrosis.
METHODS
To simulate tumor or radiation necrosis inside normal brain tissue, a 10 mm radius cylindrical inclusion was modeled inside a 150×150 mm2 rectangular prismatic domain. Typical storage modulus (2.5 kPa) and loss modulus (0.5 kPa) of brain tissue were assigned to the prismatic region. Varying tissue properties were assigned to the cylindrical region (storage modulus 0.5-5 kPa, with loss factor 0.2), mimicking the shear modulus difference between recurrent tumor and radiation necrosis. Wave propagation (30-100 Hz) was simulated in this domain, using the frequency-domain analysis in the structural mechanics module in COMSOL Multiphysics software (COMSOL, Burlington, MA). A 3D inversion method was used to estimate tissue properties.
RESULTS
In all the simulation results, the cylindrical lesion can be identified based on the modulus difference between the region and its surroundings. The lesion boundary can be delineated based on the normalized residual error (NRE) map of the fitting procedure used for modulus estimation. The estimated complex shear modulus is close to the true value when the applied frequency is or above 60 Hz. The estimate of the storage modulus is more accurate than the loss modulus. NRE is typically maximal at the boundary of the inclusion (lesion).
CONCLUSION
This study demonstrates the potential of MRE to differentiate lesions with different shear moduli. The results show the possibility of using MRE to differentiate recurrent tumor from radiation necrosis, based on the difference between their moduli, and the potential of lesion delineation based on NRE at boundaries. Physical phantom studies and ex vivo mechanical testing of the tissue will be used in the future for further verification.