This is an interesting approach that's a bit like pulse-Doppler radar.
They take a 30 fs pulse in the 800 nm region, and rectify it to get a unipolar 30-fs pulse, which has energy throughout the optical equivalent of baseband (2-50 microns). The summary article is a bit sketchy, but it looks as though they use a seldom remarked property of second harmonic generation (SHG): Just like a bridge rectifier, it generates DC as well as second harmonic. With a pulse input, the "DC" is a unipolar pulse as well.
They send that broadband IR beam through the sample, and then upconvert it back to the visible, where it's easy to detect. It looks like they have a bit of a SNR problem at the edges, but that's not too hard to understand.
Cute idea, anyway--an excellent example of the way optics resembles early radio--fancier optics can resemble more modern radio.
Cheers
Phil Hobbs