A Multispectral 3D-Endoscope for Cholesteatoma Removal

Abstract We present a stereo-multispectral endoscopic prototype using a filter-wheel to guide the removal of cholesteatoma tissue in the middle ear. An image-based method is used that combines multispectral tissue classification for the detection of tissue to be removed and 3Dreconstruction to determine its metric dimensions. The multispectral illumination used for tissue classification ranges from λ = 400 nm to λ = 500 nm with step-size of 20 nm, which results in six different narrow-band illumination modes. For classical RGB imaging and metric calculations, a broadband illumination mode is applied before and after the narrow-band illumination. The spectral information is augmented into the broadband mode using an overlay technique. The combination of multispectral imaging with stereoscopic 3D-reconstruction results in new valuable visualization of intraoperative data. This allows to generate a 3D-model of the patients anatomy highlighting the identified malicious tissue and compare the anatomical dimensions with pre-operative CT data.