The FTI routine is a maximimum likelyhood fitter which calculates the probability that the observed PMT hits came from an muon that entered and exited SNO. The fitter loops over all tubes (not just hit tubes) calculating the probability that they were hit (or not hit), and the probability for them to have observed their detected charge and time. The fitter uses a set of lookup tables (currently 108) sorted by impact parameter which contain the distribution of detected photons relative to the muon track. The lookup table is essentially a normalized pdf which contians the probability for a single photon to be detected in a given PMT. By multiplying this probability by the hypothesized number of detected photons one obtains the Poisson mean number of photons which were expected to be detected.
The fitter returns the fit entrance and direction, the impact parameter of the muon, a hypothesis for the number of photons that would have been detected in a "perfect" detector (all tubes online and no missing tubes in the neck), and a rough estimate for the energy loss per unit track length of the muon.
As of the writing of this note FTI is still under development, and the documentation is not complete. It will eventually be posted on Manhattan.