Yes, in this case it is clear the spacing is specifically with regards to electric shock, so it can be shorted as normal condition and the equipment fails the standard.
I assume that the 0.4mA was measured with the MD filter in place, meaning the "raw" leakage was about 12mA @ 30kHz. So this would fail also the 10mA leakage limit.
From experience the actual measured values are not all that stable for these systems. One reason is that the output is usually a tuned circuit with feedback and load dependence, so the voltage source is not stable. The other is that at 30kHz stray paths become significant.
One design solution for such systems is to "capacitively earth" one electrode of the transducer (the side that is closest to the patient). This capacitance then limits the leakage at 30kHz (while still providing some isolation at 60Hz).
But in reality there are a couple of reasons why this is not a significant risk:
1) the spacings in IEC 60601-1 are based on several worst case assumptions, which are unlikely to apply in a secondary circuit with regulation. In a typical ultrasound transducer design there are at least two transformers between mains and the output; pollution degree 2 applies and the spacings on a PCB with low tracking index. I would recommend looking at making sure the spacings are at least basic using IEC 60664-1 values, rather than IEC 60601-1.
2) At 30kHz, the risk of shock dramatically reduces, even more than what the MD filter compensates for. If the contact area is large (>>1mm2) also the 10mA limit is not a concern.
(by the way, I agree the standard does not exclude electronic protection, but in practice it is treated this way. If we did not, all switch mode power supplies would have to be withdrawn from the market!!).