Semi-quantitative PCR is only truly semi-quantitative if you stop the amplification somewhere between when the product is just visible on a gel and when nucleotides etc become limited and the rate of product accumulation begins to taper off significantly. That means a window of about 10 PCR cycles. If you run the reaction for too many cycles, then all of the reactions regardless of starting template concentration will go to 'completion'. You would need to get a feel for the range in which your templates fall within the parameters stated above and find an appropriate cut off point such that none of the reactions have gone to 'completion'.

Semi-quantitative PCR is only truly semi-quantitative if you stop the amplification somewhere between when the product is just visible on a gel and when nucleotides etc become limited and the rate of product accumulation begins to taper off significantly. That means a window of about 10 PCR cycles. If you run the reaction for too many cycles, then all of the reactions regardless of starting template concentration will go to 'completion'. You would need to get a feel for the range in which your templates fall within the parameters stated above and find an appropriate cut off point such that none of the reactions have gone to 'completion'.

As a rule, you probably shouldn't try to compare these techniques because of the difference in sensitivity. I would go back and repeat the RT-PCR experiments using qPCR. However, that being said, if you really need to compare them then you can flip the qPCR axis and instead of using a fluorescence threshold to generate a Ct value, you can use a cycle threshold (the same number of PCR cycles you used for your RT-PCR experiments) to calculate a fluorescence value. This can probably be compared to the quantitation of the bands from RT-PCR. But like I said, I wouldn't recommend it.