Electromagnetic tomography (EMT) is a promising and flexible technique for imaging conductivity and/or permeability because of its noncontact nature. However, the EMT inverse problem is both ill-posed and ill-conditioned because of the soft-field nature of the electromagnetic field and the limited prior information, which significantly impacts imaging quality. To overcome this challenge, an improved Cuckoo Search algorithm is introduced, incorporating a dynamic adaptive mechanism to adjust the discovery probability Pα and adaptive step size α. This modification improves the search capability, leading to higher optimization accuracy and faster convergence. Numerical simulations demonstrate that the proposed algorithm outperforms other classical techniques, particularly in terms of typical conductivity distributions. Further validation using the developed modular EMT system confirms the superior imaging performance of the method, making it a promising candidate for practical EMT applications.