The s-parameters being a small signal analysis can be easily calculated using any spice by means of AC analysis. The question is - the mos model you are using for LNA simulation is valid for RF range ?
You can refer the application note to know how to calculate s-parameter using pspice -
https://www.cadence.com/rl/Resources/application_notes/S-Parameter_Data_appnote.pdf
The two small circuits in figure 1, 4 need to be connected to your original design.
From s-parameter basic theory you can easily derive the relationship and understand what these circuits are trying to do.
I verified the method with spectre results. They match perfectly.
Suppose there is port1 and port2 and you need to find S21 and S11, connect circuit in fig-4 at port1 and fig-1 at port2.
For S12 and S22, interchange the circuits.
You can observe there is small signal (excitation) provided in fig-4, with a terminating characteristic impedance (50-ohm). At receiver side, the port is terminated separately (with 50-ohm, see examples below in appnote).
If you are connecting these circuits directly on the schematic (and not generating the hidden pins as described in appnote), you don't need R_loop1 (in fig4) and R1 (in fig1).