Multi-frequency difference method for intracranial hemorrhage detection by magnetic induction tomography

Abstract

Objective: Frequency-difference magnetic induction tomography (fdMIT) is a promising technique for the continuous monitoring of intracranial hemorrhage. The dual-frequency difference MIT (dfdMIT) will sometimes produce artefacts in the reconstructed images. To overcome this problem, a multi-frequency difference MIT (mfdMIT) method is proposed. Approach: To test the performance of this algorithm, 2D head models comprising six tissue types with different hemorrhage sizes were simulated, and images of the hemorrhage were reconstructed by mfdMIT and dfdMIT. Imaging errors and correlation coefficients were calculated with different levels of added phase noise and conductivity noise. Main results: A hemorrhage with a diameter of 20 mm can be reconstructed with 20 dB noise by the mfdMIT method, and the imaging error is reduced by about 60%. The results show that the conductivity of brain tissues changing with frequency will cause artefacts on the reconstructed images of the hemorrhage by dfdMIT, but the artefacts can be effectively suppressed with mfdMIT by separating the hemorrhage from other brain tissues. Significance: The mfdMIT method increases the reconstructed image precision, which promotes the development of medical imaging by MIT.