I haven't tried it at all, but I suspect that you could use linear algebra.
If you turn each line-segment into a vector, then this layout as we see would exist in
. From this, we can see that QP is the result of QS cross QT. If you define Q as the origin of our vector space, you can then set QS as (a, b, c), QT as (d, e, f) and QP as (0, 0, g), you could then use the varying vectors/line segments to find a, b, c, d, e, f and g, and then you would be able to find the angle SQT, which would mean you can use the cosine rule to find ST or simply find the length of the vector ST from basic vector addition.
Might not have enough information though - this is all conjecture, I haven't actually tried to evaluate it, nor do I have the time to try.