tl;dr: Link to paper & data at the bottom
Over the past few months, Brandon Hensley, Olivier Doré, and myself have revisited the correlation between interstellar reddening by dust grains and the atomic neutral hydrogen column density. The basic idea is that HI and dust grains are well-mixed in the ISM and thus that HI can also be used as a measure of E(B-V), which has already been demonstrated in the late 70s.
We find an excellent agreement between the two ISM tracers, down to the mmag level, for large (~40%) areas of the sky. In the parts of the sky that do not agree quite so well, we find that our corrections to SFD are highly correlated with the SFD corrections published by Peek & Graves (2010). The latter is a low-resolution correction, based on passively-evolving galaxies, and is thus completely independent from the HI-based measurement. We argue that in the low column density regime which is ultimately relevant for cosmology, the HI-based E(B-V) map is preferable. It has been known for quite some time that distant, unresolved FIR galaxies can bias the SFD map slightly (Yahata et al. 2007).
More information on the paper, the map, and accompanying tools can be found here.