I think you should be able to achieve an Isp approaching that of this rocket with a solid core nuclear reactor, without radiators, although at much lower T/W ratio.<p>The idea would be to not just dump hydrogen into the reactor to heat it, but gradually warm that hydrogen, extracting as much power as one could along the way. At the end, this power would be used to superheat the hydrogen after it went through the reactor (by some sort of electrical heating), to a temperature greater than the reactor's temperature limit. Alternately, the exhaust stream could be further accelerated by some sort of MHD afterburner.<p>The observation here is that a nuclear rocket is not energy limited, but rather is entropy limited: the exhaust can only carry away so much entropy. So, the goal is to make the engine as internally efficient as possible, with as little thermodynamic irreversibility as possible.