Fast detection and reconstruction of defect geometry is one of the main goals of ECT. Inversion of ECT data usually requires high-computational time and resources, hampering the application in on-line or in-situ inspections. In [1-2], it was hypothesized that defects’ patterns in 2D ECT images can be modelled as a 2D convolution between defects’ geometrical shape and a Hermite-Gauss (HG) mode function, which depends on the sensor. By knowing the mode function of a given sensor, it is possible to exploit image deconvolution to increase the SNR of detection and for a fast estimate of the defect shape and dimensions. To do a further step in this direction, in this work various sensors, excitation strategies and analysis methods were used to test a benchmark sample containing known defects. It was confirmed that for all the various configurations, 2D images in time- or frequency- domain are well described by the HG hypothesis.