─ Industrial structure are exposed to microstructural changes caused by fatigue cracking, corrosion and thermal aging. Generally, a hidden crack is very dangerous because it is difficult to detect by Non-Destructive Evaluation (NDE) techniques. This paper presents a new approach to estimate the hidden cracks dimensions inside a stainless steel plate based on the EMAT signal. The received signal by EMAT is simulated using the Finite Element Method (FEM). Then, the identification of the hidden crack sizes is performed via the combination of two techniques; the first one is the Time-of-Flight (ToF) technique which was applied to estimate the crack height by the evaluation of the difference between the ToF of the healthy form and the defective form. Then, the crack width is estimated by the solution of the inverse problem from the received signal based on a meta-heuristic algorithm called Teaching learning Based optimization (TLBO). The obtained results illustrate the sensitivity of the EMAT sensor to the variation of the crack sizes. Moreover, the quantitative evaluation of the cracks dimensions, show clearly the efficiency and reliability of the adopted approache.