Task-specific multispectral band selection

TL;DR: The investigated domain adaptation technique, which only requires unannotated in vivo measurements yielded better results than the pure in silico band selection method, and could guide development of novel and fast multispectral imaging systems suited for interventional use without relying on complex hardware setups or manually labeled data.

Abstract: Multispectral imaging provides valuable information on tissue composition such as hemoglobin oxygen saturation. However, the real-time application of this technique in interventional medicine can be challenging because the acquisition time of large amount of hyperspectral data with hundreds of bands can be long. This challenge can partially be addressed by choosing a discriminative subset of bands. The band selection methods proposed to date are mainly based on the availability of often hard to obtain reference measurements. We address this bottleneck with a new approach to band selection that leverages highly accurate Monte Carlo simulations. We hypothesize that a so chosen small subset of bands can reproduce or even improve upon the results of a quasi continuous spectral measurement. We further investigate whether novel domain adaptation techniques can address the inevitable domain shift stemming from the use of simulations. Initial results based on in silico and in vivo experiments suggest that 10-20 bands are sufficient to closely reproduce results from 101 band spectral measurement in the 500-700nm range, depending on the experiment. The investigated domain adaptation technique, which only requires unannotated in vivo measurements yielded better results than the pure in silico band selection method. Overall, our method could guide development of novel and fast multispectral imaging systems suited for interventional use without relying on complex hardware setups or manually labeled data.