The authors describe a fully automated segmentation approach for mandibular bone and brain segmentation in CT data used in radiotherapy planning based on a joint segmentation-registration framework.

Radiation therapy planning of head and neck cancer.

The average volume overlap with manually delineatied ground truth by this approach is 74% for the mandible and 72% for the brainstem.

The paper is well written with clear explanation of the applied method. Some minor remarks to consider are below.

Introduction, par. 3

Please mention which H&N structures (mandible and brainstem representing a bony structure and a soft tissue organ) are in focus of this work.

p.6, par. 3.

Is brainstem the soft tissue structure of interest in this case? Is the similarity metric computed only in the area of brainstem? Please clarify.

p. 6, par. 4.

The meaning of the evaluation criterion of HD larger than 3mm should be explained in the text.

Evaluation results

Please mention that dataset no. 13 was excluded from the brainstem evaluation due to truncated field of view.