Applying Frequency-Wavelet Domain Deconvolution to in Vivo Terahertz Measurements of Skin Conditions

Abstract

Eczema and psoriasis are two of the most prevalent skin conditions in the United Kingdom. Typically, diagnosis and treatment assessment are evaluated based on qualitative measures such as skin appearance and patient feedback. Thus, there is a need for a robust and quantitative method of assessing treatment efficacy and monitoring improvements in skin condition. Terahertz (THz) sensing is a promising candidate, owing to its nonionising properties and high water sensitivity. However, THz sensing is heavily reliant upon signal processing techniques, in particular deconvolution methods used to extract skin-dependent parameters. Conventionally, double Gaussian deconvolution has been used, however this has been shown to oversmooth data and cause unnecessary signal suppression. In this study, an alternative approach, known as frequency-wavelet domain deconvolution, has been applied to in vivo THz skin measurements obtained in a clinical study. It is shown that this deconvolution method significantly enhances the usable signal, and improves the accuracy of the extracted impulse response function.