Nonnegative matrix factorization: a blind spectra separation method for in vivo fluorescent optical imaging

Anne-Sophie Montcuquet (1, 2) , Lionel Herve (2) , Fabrice Navarro (2) , Jean-Marc Dinten (2) , Jerome I. Mars (1)

1 GIPSA-SIGMAPHY - GIPSA - Signal Images Physique
2 CEA-LETI - Commissariat à l'énergie atomique et aux énergies alternatives - Laboratoire d'Electronique et de Technologie de l'Information

Fluorescence imaging in diffusive media is an emerging imaging modality for medical applications that uses injected fluorescent markers that bind to specific targets, e.g., carcinoma. The region of interest is illuminated with near-IR light and the emitted back fluorescence is analyzed to localize the fluorescence sources. To investigate a thick medium, as the fluorescence signal decreases with the light travel distance, any disturbing signal, such as biological tissues intrinsic fluorescence (called autofluorescence) is a limiting factor. Several specific markers may also be simultaneously injected to bind to different molecules, and one may want to isolate each specific fluorescent signal from the others. To remove the unwanted fluorescence contributions or separate different specific markers, a spectroscopic approach is explored. The nonnegative matrix factorization (NMF) is the blind positive source separation method we chose. We run an original regularized NMF algorithm we developed on experimental data, and successfully obtain separated in vivo fluorescence spectra.

Domaines: Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Anne-Sophie Montcuquet, Lionel Herve, Fabrice Navarro, Jean-Marc Dinten, Jerome I. Mars. Nonnegative matrix factorization: a blind spectra separation method for in vivo fluorescent optical imaging. Journal of Biomedical Optics, 2010, 15 (5), pp.0560091-14. ⟨10.1117/1.3491796⟩. ⟨hal-00583030⟩