Segmentation of seventy-one anatomical structures necessary for the evaluation of guideline-conform clinical target volumes in head and neck cancers

Alexandra Walter, Goran Stanic, Philipp Hoegen, Sebastian Adeberg, Oliver Jaekel, Martin Frank, Kristina Giske

Keywords: Expert guidelines, automatic segmentation, anatomical structures, multi-label segmentation, target volume delineation, head and neck cancer

TL;DR: This paper presents the results of automatic segmentation for seventy-one anatomical structures that are necessary for the evaluation of guideline-conform clinical target volumes in head and neck cancers.

Abstract:

Expert guidelines standardize the delineation of the clinical target volume by giving neighboring anatomical structures as boundaries. In this research, we have automated the segmentation of seventy-one most mentioned anatomical structures in the expert guidelines. For most structures there are no segmentation accuracies found in the literature. For others, the DICE improves in our research. The sDICE with a 2 mm tolerance shows clinical acceptance for most structures. Overall, we set benchmarks for several structures of the
head and neck anatomy. With these segmentations the guideline-conformance of clinical target volumes can be measured.