Abstract

Laser angioplasty suffers from a high incidence of vessel perforation. Laser-induced fluorescence spectroscopy can be used to discriminate between normal and abnormal (atheromatous) tissue and thus guide the delivery of the ablative laser energy. An important issue is how to process the fluorescence spectra to obtain an optimally correct classification of tissue type. An approach based on a maximum likelihood criterion is presented. The discriminant function that results from application of this approach is determined using a large training set of spectra and evaluated on a subsequent validation set. The performance of the method is discussed.

© 1987 Optical Society of America