A plasma boundary reconstruction method based on optical images was introduced by Hommen et al.. Consistency of the boundary with the boundary from magnetic measurements was demonstrated for a few cases, but systematic validation over a wide range of plasma parameters was lacking. Here we present a systematic comparison of the EFIT (magnetic) and OFIT (optical)
boundaries for more than two hundred MAST (Mega Ampere Spherical Tokamak) discharges. For this comparison, the OFIT concept had to be implemented in the MAST data analysis chain and needed to be improved on a number of points. A dimensionless figure of merit has been introduced comparing
the non-overlapping regions with the overlapping regions. With this criterion, overall discrepancies of about 6-8% were found between the optical and magnetic boundary. These dimensionless discrepancies translate to overall distances between the EFIT and OFIT boundaries of about 2-3 cm. Upper-limits on the error of OFIT, of 2:7 cm for the DND case and 4:1 cm for the SND case, were found. No (strong) correlation was found with plasma parameters such as normalized pressure ß or self induction li. We show that a significant part of the discrepancy is likely caused by faults in the EFIT reconstruction. This leads to the conclusion that the OFIT technique, if implemented and used correctly, is usable for reliable real-time detection of plasma position and shape, with an overall error less than 3-4 cm.