i would not do it that way.
you show an elliptical lowpass filter. If you make use an open circuited stub to approximate the L-C shunt resonator, then you simply set the electrical length of the stub to be quarterwave where the L-C element resonated in frequency. You choose the impedance of the stub transmission line to approximate the reactance slope of the lowpass prototyupe L-C network at the frequency near the cuttoff of the filter.
the series inductor elements....you just pick the highest impedance line you can make on the microstrip (say that is 90 ohms). you then make the length of that narrow width transmission line such that the transmission line looks like the reactance of the series inductor near the freequency of cuttoff for the lowpass filter.
after you put all the transmission line elements together, and account for T junctions and width step discontinuitites, you might need to tweek the lengths and widths just a little to get the ripple down low and rejection where you want it. BTW, you will NOT be able to get as good a rejection with the transmission line model as you could with the ideal lumped element model.