At the beginning of a machine operation, an assessment of the intrinsic error fields, spurious magnetic field perturbations which can affect plasma dynamics, is often carried out by executing the compass scan method [Scoville J.T. et al. Nucl. Fusion 43 250 (2003)]. This method relies on the application of 3D magnetic fields with various phases, induced by EF correction coils, to trigger a locked mode. The instant of locked mode onset allows the identification of the amplitude and phase of the intrinsic error field, from which the empirical correction currents for its minimization can be deduced. The presence of a locked mode needs to be carefully monitored during this study because of the potential disruptive mode behavior, especially in devices which can tolerate a maximum number of disruptions, as in SPARC and in ITER. A novel method, the so-called non-disruptive compass scan method [Paz-Soldan C. et al., Nuclear Fusion 54 (2014) 073013], avoids the disruption risk, as the name recalls, via magnetic island healing, i.e. stabilizing the locked mode. The magnetic island healing is achieved by switching off the error field correction coil current during the execution of the compass scan and asynchronously by increasing the plasma density. The crucial point of this new method is the detection of the locked mode to initiate the EFCC-density control actions. In this work, the locked mode detector adopted during non-disruptive compass scan test at JET is presented, together with brand-new locked mode metrics, which take into account the actual poloidal deformation due to a locked mode and a class of MHD instabilities, named Beta Alfve & PRIME;n Eigenmodes, that appear in the Mirnov signal in concomitance to the locked mode. The use of multiple metrics for locked mode detection during the execution of the non-disruptive compass scan increases the fidelity of the real-time control system to pinpoint the event, compensating possible magnetic probe failure, and initiate the control sequences to heal the magnetic island.