Best means maximum thrust, safest, easy to maintain. And it's not opinion based because I'm looking for a way that's theoretically feasible and may be practical in a few years. However, if you have an idea of your own (howsoever wacky it might be, because you never know if that's the best way in the future), please tell me about it.
Exploding thermonuclear bombs behind a big, thick plate, with the payload on giant shock absorbers behind that, referred to as Project Orion, is practical today, and has been for decades.
There is no other practical way to generate net energy from fusion, and there won't be for decades.
The current situation about the fusion
Today, there is no working fusion reactor today which could produce energy. The current seemingly best prototype is the ITER reactor, which still won't produce energy for the European energy network, but it will produce at least more energy (from fusion) as it takes (for maintaining plasma). It is currently under production, the first plasma reaction is planned for 2025 (which is already delay to the originally planned 2018).
After the ITER demonstrated the capability to produce fusion energy, the next planned reactor (DEMO, existing currently as preliminary plans) will already produce electricity for the continental power system. This is estimated for 2040.
Note, the ITER tokamak weighs 23000 tons, and there is no reason to think that the DEMO will be lighter. Compare this to the ISS, which is the most massive thing currently in space, with its 450 tons.
Thus, after the DEMO is ready and working, probably it will require further decades to miniaturize the reactor and making it space-capable.
Note, the only currently working industrial fusion doesn't fuse Hydrogen. It uses Lithium and Deuterium, produces partially Tritium from them, and then uses this Tritium to create fusion.
Current situation about the nuclear-capable space propulsion
The mass of the decay products of the nuclear energy is very small, compared to the chemical rockets. This rules out the current chemical rocket-like solutions (although there are some ideas even for that).
Having the very high energy level and the very small fuel mass, ion drives are the most obvious choice. Between them, VASIMR has the most fruitful prototype.
VASIMR is essentially a very effective ion engine. If it gets (electrical) energy and it has fuel, it can work with an effectivity around 70%. It can get its electric input from batteries, from solar power or from a fusion reactor.
There is currently a working VASIMR prototype which could reduce the orbital boost costs of the ISS to the 5% of the today, and there was a plan to install it on the ISS in 2015. The NASA stepped back from it in the last moment, on unclear reasons. Their this decision makes probable, that the ISS will be simply crashed into the Pacific sometimes in the 2020s, for example on cost reasons.
The company behind the VASIMR still works for the NASA, although it is dubious how can they do any useful, considering that currently there is no plan to use VASIMR anywhere.
A fusion reactor, converting Lithium and Deuterium to Helium4, Hydrogen and energy, driving VASIMR drives seem the most realistic outcome.
It couldn't run before 2060 even in the best scenario.
In the latest decades, the developments in the space exploration, the actually realized scenarios were more near to the worst.
Hard to say if it's best, but it's certainly both very promising, very efficient - both fantastic specific impulse, and very little fuel wasted (unlike most fusion based propulsions where only small percentage of the fuel undergoes fusion, here nearly all deuterium and tritium is used up; resulting alpha particles being the propellant). Also, unfortunately fantastically expensive: Muon-catalyzed fusion.
Unlike proposals like Project Orion, or ITER, it scales very well, working equally well for nanosatelites as small as 6U cubesats and craft as big as generation ships (...although the cost scales linearly; for bigger craft other proposals may be more economically viable).
The one bad problem is it depends on anti-matter as catalyst of the fusion. Unlike in annihilation, where one positron + one electron produce two photons, the end, here one positron can catalyze fusion of several hundred and more of deuterium or tritium atoms, releasing much more energy than its own annihilation would produce. That makes requirements for the amount of anti-matter (or more specifically, an isotope of a common element, that produces positrons on radioactive decay) needed to run quite small. Quite small still doesn't mean insignificant though, and production of these isotopes is awfully expensive.
Let me add what way of converting the fusion energy into propulsion would be objectively best, disregarding the source of the fusion and the current technology readiness level:
As two atoms/particles undergo fusion (e.g. deuterium+tritium turning into helium), their products are extremely hot. If you are able to focus these products into an exhaust nozzle and just eject them, you're getting the optimal propulsion - least losses, the highest exhaust momentum possible, shedding inert mass most rapidly. Anything else will be more lossy.
Now, how to direct the fusion products into the nozzle (and not vent all the substrates simultaneously) is one challenge on par with achieving sustainable fusion.
BTW, the fusion doesn't need to be energy-positive! You can deliver a surplus energy, say, beaming it, lose it on causing fusion, produce less through fusion than you lost, and still if you can eject the helium in a coherent 'exhaust flame' you're good - Fusion is a fantastic way of accelerating particles to huge speeds, which is an essential last step of any reaction drive, completely regardless of the source of power!
Theoretically whatever might be done to make controlled fusion work in a reactor could be adapted to a rocket. One fusion reactor approach is to detonate micro pellets of deuterium/ tritium in a cross fire of lasers. If the reactor is open to space on one end, the resulting plasma would exit out and create thrust. It could be diluted with additional reaction mass to increase thrust and cool the chamber walls. A fusion reactor in a magnetic bottle could become a rocket by bleeding some plasma out a magnetic nozzle. Again adding reaction mass to the plasma would increase thrust.
Arguably, this has been done by Dawn, which uses a solar powered ion drive. So, the answer used in that case was to assemble a sufficient mass of hydrogen that it starts to fuse simply due to gravitational confinement. Then catch (some of) the resulting radiant energy and use it to power a Xenon ion drive. Helpfully for the Dawn team, the first part of this had already been achieved.