A Novel Lebesgue-Space Inversion Approach for Microwave Stroke Imaging

Brain stroke is a pathology affecting a significant number of people worldwide. A fast diagnosis is required to administer the correct therapy and frequent monitoring is usually necessary during follow-up. Microwave imaging represents a promising technology able to address such needs and that may complement gold-standard systems. In this framework, a novel Lebesgue-space approach for microwave imaging of stroke is presented in this paper. The developed method combines an inexact-Newton inversion technique in variable-exponent Lebesgue spaces with a data-driven regularization term. In this way, a-priori information about the configuration under test, e.g., the head structure, can be exploited and the convergence of the method can be enhanced. The method is preliminarily evaluated by means of numerical simulations involving realistic head phantoms. The obtained results confirm that the developed approach allows identifying and characterizing the stroke with a good accuracy.