The orthotropic steel decks (OSDs) are widely used in long-span bridges due to their lightweight attributes. However, they face challenges with fatigue cracks. A common type of fatigue crack in orthotropic steel decks is the root-deck crack, which starts propagating inside the U-rib between the weld and the steel plate. Since subsurface’s location, this subsurface crack is difficult to detect in its early stages. Therefore, an efficient detection method and development are necessary to prevent major damage to the deck and entire structural system. This study introduces an eddy current technique, combined with a newly proposed signal processing method and an optimal probe configuration, to improve and enhance the detection of subsurface cracks through numerical analysis. The analysis involves using dynamic electromagnetic field analysis to simulate a Pulse Eddy Current input and output signal. The newly proposed signal processing is applied with the results of analysis to demonstrate its ease of understanding and its performance under challenging conditions. Additionally, the study includes an optimization of the probe’s configuration. This optimization considers the probe’s performance comparison with conventional probes, its performance in off-center positions, and the effect of the steel plate’s edge on the probe’s performance. The results show that the newly proposed signal processing method exhibits robust performance, as it is a relative analysis capable of handling noise and various amplitudes of input current. Furthermore, with this new signal processing approach, the study proposes an optimal probe configuration, the cup-core probe, which improves the output signal of Eddy Current Technique and functions effectively under challenging conditions.