Segmentation Using Near-Infrared Imaging: An Application to Skin De-oxygenation

The paper presents processing of Near-Infrared Spectrum (NIRS) images of skin wounding rat model by using different image processing tools. The Near-Infrared Spectrum (NIRS) images provide information on biological tissue oxygenation. We had tested the possibility of detecting the lesions made using traditional lancet and acusector. In the obtained NIRS images, background noise limiting the segmentation process were removed in the transformed domain. We used the thresholding (grayscale morphology) of wavelet coefficients to remove background. Segmentation of the interest areas in two different scenarios were obtained by means of thresholding based region growing method. The preliminary results obtained are quite encouraging and could be significant for real-time medical applications using infrared image segmentation. Furthermore, the obtained results supports the view of using NIRS for the evaluation of skin disease and microcirculatory dysfunction.