Quantitative measurements of soft tissue structures using image-features in navigated endoscopy

Below we present a new method to obtain quantitative measurements in medical images of a tracked video endoscope. Such quantitative measurements can be critical biomarkers during endoscopic diagnosis, e.g. for the classification of tumors and polyps. Size-based classifications are often achieved by mere visual estimation which can be imprecise and subjective. To this day established systems of quantitative size measurement are restricted to specific anatomical structures and require additional tools next to the endoscope. We present a system of quantitative measurement during an endoscopic procedure based on a virtual model by utilizing methods from computer vision as well as structure from motion and combining those with an optical tracking system in a novel way. Our experiments show that quantitative measurements of anatomic structures can be reliably attained by our system with an accuracy which exceeds the reported accuracy of visual estimation.