Abstract

We present a novel method, space-space MUSIC (Multiple Signal Classification), to localize three-dimensionally focal fluorophore-tagged lesions activated subsequently by different laser source positions from multi-sensor fluorescence data obtained from a single measurement plane. Matches between a signal subspace derived from the measured data and data from model spots allow 3D determination of the centers-of-gravity of fluorescence regions. Simulated spots in bounded, inhomogeneous media could be localized accurately. The algorithm has shown to be robust against patient-dependent parameters, such as optical background parameters. The algorithm does also not consider medium boundaries.

© 2003 SPIE