Non-invasive tumor detection using spectrally-resolved in vivo imaging

Genady Kostenich, Sol Kimel, Zvi Malik, Arie Orenstein

Research output: Contribution to journalArticlepeer-review

1 Scopus citations


A novel spectral image-analysis system was used for tumor fluorescence and reflectance imaging in an animal model and in patients. Transcutaneous fluorescence imaging was carried out on Balb/c mice bearing subcutaneous C26 colon carcinoma after intraperitoneal (i.p.) administration of 5-aminolevulinic acid (ALA), a metabolic precursor of protoporphyrin-IX (PP), and of a novel photosensitizer tetrahydroporphyrin (THP). Tumors were clearly observable by fluorescence detection using green light excitation. Tumor versus normal tissue uptake of the photosensitizing agents was determined by monitoring fluorescence intensity. Maximal PP accumulation in tumor was observed 3 h after i.p. injection of ALA, whereas THP showed selective accumulation in tumor 24 h after administration. Reflectance spectroscopy was employed to study pigmented human skin lesions (nevus, pigmented BCC and pigmented melanoma). In the near-infrared region (800-880 nm) pigmented BCC and melanoma exhibited a differently shaped reflectance spectrum compared to normal skin and nevus. Spatially and spectrally resolved imaging, in combination with mathematical algorithms (such as normalization, spectral similarity mapping and division) allowed unambiguous detection of malignancies. Optical biopsy results from a total of 51 patients showed 45 benign nevi, 3 pigmented BCC and 3 malignant melanomas, as confirmed by histology.

Original languageEnglish
Pages (from-to)244-251
Number of pages8
JournalProceedings of SPIE - The International Society for Optical Engineering
StatePublished - 2000


  • Benign nevus
  • Fourier transform spectroscopy
  • Melanoma
  • Near-infrared reflectance
  • Pigmented basal cell carcinoma


Dive into the research topics of 'Non-invasive tumor detection using spectrally-resolved in vivo imaging'. Together they form a unique fingerprint.

Cite this