Yes -- difference being, at the ~30kHz a typical ultrasonicator operates at, the waves seem to propagate quite nicely. I suppose to be fair, even at some MHz they do, else nebulizers would have a much harder time working. (Hmm, want to say their beams dissipate quite quickly, say in the span of a few mm; though whether that's due to simple spreading and reflection, I don't know.) But yeah, cavitation should do the trick.
Hm, I wonder if it would be worth adding lowpass filter sections, ahead/behind the active section. A ~1mm tube of fluid will act as a waveguide, carrying energy away from the source and thus reducing the amount of cavitation produced. And not necessarily dissipating that energy as heating in the fluid, it might go mostly into the sidewalls or something instead (which will still conduct back into the fluid so that's okay, as long as it's low enough power density that it behaves itself). Maybe it's not enough to matter? I wonder what such a section would look like, anyway; you don't have much opportunity to change the I.D. of the tubing, assuming some requirements for flow; maybe it's enough to use spirals of soft (silicone?) tubing, and just let the waves disperse. Oh, also a lot of energy can be carried on hard metal parts; soft tubing connections should serve as shock absorbers, again tending to keep the energy in the active region. As opposed to, say it's a benchtop unit with two hose barbs on the front, you might not want metal tubes going back from them to the heating cell.
Heh, one interesting side effect of this, I would think: as the fluid is heated, some cavities will not collapse, but stay inflated by some amount of dissolved gas that's come out. Like when you raise tap water to near boiling, in a clean vessel, at first it's bubbling slowly, but add a nucleation source and a fog of fine bubbles are released. In this case, that nucleation has already happened, so it should be about as degassed as it's going to get, at whatever temperature it's been raised to. Put more succinctly: the state of dissolved gas should be closer to equilibrium.
Which also means, with a ready source of nucleation, the threshold of critical heat flux should be lower, for example.
And I wonder what effect a slightly frothy load will have. The density will be lower, so a higher mechanical impedance. Cavitation won't be entirely by spontaneous nucleation and collapse, but by (mostly-)adiabatic cycling of the bubbles, but that should still transfer plenty of heat.
Y'know, probably it's a lot like induction heating, where the Q factor might not be all that low, it's not like wiring up a resistor, you pretty much always have to deal with the reactance -- but it's also not crazy, that reactance can be tuned out easily enough, and there you go. At least up to a modest variation in load characteristics, which may require adaptive tuning or frequency control to account for, depends. Biggest difference is you're doing it in space, all the propagating modes matter. It's ultimately a one-port to the circuit, probably the impedance doesn't change much, easy enough to drive; what would change more is the spacial distribution, or coupling efficiency to how much cavitation, that sort of thing I would think.
Tim