Towards accurate target delineation for head and neck cancer

To benefit from these high precision techniques, an accurate delineation of the target is essential. In this thesis, imaging modalities for Gross Tumor Volume (GTV) delineation for laryngeal and hypopharyngeal tumors are validated by using histopathology as a gold standard. The need for this research stems from a general problem in radiotherapy compromising the accuracy of the radiation treatment: variation among observers in delineation of the target. This problem was outlined by an interobserver study, which resulted in high variation between observers for GTV delineation for supraglottic laryngeal carcinoma on CT-images and CT-images combined with MR-images. When interobserver variation is high, precision is low. What about the accuracy of the delineation, and how to increase this accuracy? In this thesis, images on various imaging modalities were compared to histopathology of on average 800 H&E-sections of 27 surgical laryngeal specimens. The specimens and H&E-sections were reconstructed in 3D to enable comparison to the in-vivo images accomplished before surgery. Several steps were undertaken to optimize the method used for validation. The precision of the gold standard was investigated first. Guidelines for GTV delineation on MR-images were developed to increase the accuracy of the delineation. Overestimation and underestimation of tumor tissue were determined before and after the appliance of the guidelines. The guidelines halved the overestimated volume of the GTV. CTV is the volume of tissue that contains a demonstrable GTV and/or subclinical malignant disease with a certain probability of occurrence considered relevant for therapy. This microscopic spread is not visible on images. However, the degree of visibility varies between imaging modalities depending on their specific qualities. The size of the CTV margin is merely historically determined and varies between institutes. A distance analysis was conducted based on underestimation of the histopathologically determined tumors for CT, MRI and FDG-PET. These distances were used to develop modality specific CTV margins. GTVs with addition of these margins resulted in CTVs that are halved in volume compared to CTVs that were derived from GTVs with the clinically used CTV margin of 10 mm. The addition of modality specific CTV margins would lead to smaller volumes to be treated with high dose radiation and consequently to a decrease in radiation to non-tumorous tissues resulting in less damage to healthy tissue