Spatial dispersion of action potential duration (APD) restitution (APDR) due to electrophysiological heterogeneities in the heart, has been suggested to act as a potent arrhythmogenic substrate. In this work, we evaluate a method aimed at quantifying APDR slope dispersion (Δα) from the surface electrocardiogram (ECG). A 2D human ventricular in silico tissue preparation is used, from which APDR slope dispersion (Δα^SIM) at tissue level is computed, and compared with simulated estimates calculated from pseudo-ECGs (Δ̂α^pECG). We show that Δ̂α^pECG values in our simulations are in all cases within the range of clinical values measured from tilt-test recordings. Then, we use the validated tissue model to relate the Δ̂α^pECG estimate to measurements of Δα^SIM, and we prove that the mean error is below 5%. We conclude that the proposed Δ̂α^pECG index provides valuable estimates of APDR dispersion with the advantage of being able to be measured non-invasively from the ECG.