6533b7d3fe1ef96bd126080e

RESEARCH PRODUCT

Application of principal component analysis to multispectral imaging data for evaluation of pigmented skin lesions

Dainis JakovelsJanis SpigulisIlona KuzminaIlze Lihacova

subject

Pathologymedicine.medical_specialtyintegumentary systembusiness.industryWavelength rangeMelanomaMultispectral imagemedicine.diseaseDermatologyPrincipal component analysismedicinePigmented Nevussense organsSkin cancerPigmented skinSkin melanomabusinessneoplasms

description

Non-invasive and fast primary diagnostics of pigmented skin lesions is required due to frequent incidence of skin cancer – melanoma. Diagnostic potential of principal component analysis (PCA) for distant skin melanoma recognition is discussed. Processing of the measured clinical multi-spectral images (31 melanomas and 94 nonmalignant pigmented lesions) in the wavelength range of 450-950 nm by means of PCA resulted in 87 % sensitivity and 78 % specificity for separation between malignant melanomas and pigmented nevi.

yearjournalcountryeditionlanguage
2013-11-18SPIE Proceedings
https://doi.org/10.1117/12.2044383