Several metrics have been proposed in ultrasound pulse-echo imaging to quantify local coherence, with the coherence factor being the most prominent example. However, the coherence factor does not differentiate losses of coherence caused by variations of speed of sound within tissues and other factors, particularly multiple scattering. In this work, we propose a metric designed to estimate local coherence independently of speed-of-sound variations in the insonified medium. Our method relies on a recently introduced adaptive beamformer for plane-wave imaging based on the windowed Radon transform of images reconstructed with delay-and-sum. We demonstrate with simulated data that the proposed metric is only very weakly affected by speed-of-sound variations compared to the coherence factor, and we present promising preliminary in-vivo results.