Multifrequency Difference Magnetic Induction Tomography With Robust Independent Component Analysis for Hemorrhage Detection

In the detection of intracranial hemorrhage by magnetic induction tomography (MIT), frequency difference (fd) imaging can reconstruct the conductivity distribution of the hemorrhage without the reference measurements before bleeding. However, the influence of background head tissues deteriorates the imaging quality. To overcome the difficulties of hemorrhage detection by MIT, in this study, we propose a multifrequency difference with robust independent component analysis (mfd-RICA) method. Based on the multifrequency response model of head tissues in MIT, the phase shift of hemorrhage is extracted from the measurements by the RICA and used for conductivity reconstruction. To evaluate the performance of mfd-RICA, both numerical simulation and phantom experiment were carried out. The results show that mfd-RICA has high extraction accuracy of the hemorrhage phase shift, which can effectively suppress the influence of background tissues and improve the accuracy of hemorrhage reconstruction. Moreover, mfd-RICA is proved to be robust to the noisy data, which facilitates the accurate detection of intracranial hemorrhage. Potentially, this method can enable the detection of acute intracranial hemorrhage with portable MIT devices.