Characterization of healthy skin using near infrared spectroscopy and skin impedance

Abstract

Near infrared spectroscopy (NIR) and skin impedance (IMP) spectroscopy are two methods suggested for diagnoses of diseases inducing adverse effects in skin. The reproducibility of these methods and their potential value in non-invasive diagnostics were investigated. Measurements were performed in vivo on healthy skin at five anatomic body sites on eight young women. partial least squares discriminant analysis showed that both methods were useful for classification of the skin characteristics at the sites. Inter-individually the NIR model gave 100% correct classification while the IMP model provided 92%. Intra-individually the NIR model gave 88% correct classification whereas the IMP model did not provide any useful classification. The correct classification was increased to 93% when both datasets were combined, which demonstrates the value of adding information. Partial least squares discriminant analysis gave 72% correct predictions of skin sites while the combined model slightly improved to 73%.