Towards a Surgical Tool Using Hyperspectral Imagery as Visual Aid

This paper presents a new application to exploit the capabilities of hyperspectral imagery as a visual supporting tool during surgeries. In order to enhance the visualization of regions affected with spilt blood, a hyperspectral imaging technique has been developed. We propose a neural network approach to nonlinearly combine the wavelengths of the spectrum in order to reduce the effect of spilt blood in an image and produce a clearer image of the concealed area. Preliminary experimental results guinea pigs are presented to validate the method. A sketch of a four-band multi-spectral camera, tuned in wavelengths identified using the learning algorithm, is also presented.