2017 Volume 55Annual Issue 3AM-Abstract Pages 181
Penetration depth of the light into human tissue depends on its wavelength because of optical characteristics. Narrow-band imaging has a potential to provide a variety of texture images depending on the penetration depth that can be useful for tissue tracking, while its imaging has been studied intensively for lesion visualization. This study aimed at developing a robust and dense deformation tracking method of organs for image-guided surgical intervention using narrow-band imaging. We proposed a novel template matching method with multiple layers consisting of different spectral narrow-band images as a coarse-to-fine approach. We analyzed and found the relationship between light wavelengths and spatial frequencies of images. We also conducted an experiment with a chicken liver to evaluate accuracy of matching and tracking. The results showed that the proposed method was more accurate compared with the methods with single spectral images. We also discussed the characteristics of the proposed method for tracking.
