how to attach the high current mosfet to the bus bars
It would probably be a good idea to stick a thermocouple on it to measure the temperature.
Surgeon general warning: do not cook with lead.
Note IRFU7540 in your schematic is not the proper MOSFET for the job. You need one with tiny RdsON, that can take the current both through the chip and source leads, in a package without bondwires. The NXP video uses a LFPAK FET which has a clever source leads arrangements and no bondwires.
But you should really use a package with bolts, as Jeroen says.
Since it's a NMOS you need to drive the gate above the +48V supply, so you need a boosted supply voltage. Your choice, canned isolated DC-DC, charge pump...
It should be able to switch off very fast in case there is a short, because if there is a short, current will rise to many thousands of amps, and during switch-off the FET will have the full VxI dissipation (ie, 48V x thousands of amps) so you don't have much time before it blows.
Ideally you want to exploit the wiring inductance to limit the rate of rise of current during a short, so the entire overcurrent protection chain, from current measurement device, comparator, etc, has to react extremely quickly. If you got 100nH of total inductance in your wiring, in 200ns on a short the current will only rise by 96 Amps. So if you have a very fast protection circuit you can turn off the FET before current rises to dangerous levels. On the other hand if your protection circuit takes 2µs to react, by then current will have reached a thousand amps, and FET turn-off losses will thus be 100x more.
If you use software to implement your short circuit protection, you will find bits of MOSFET embedded in the ceiling when it blows. It has to be hardware, it has to be fast and foolproof, and it has to work even if the CPU and all peripherals are frozen by the debugger.