The Study Of The Landmarks For Image Registration In Upper And Mid-thoracic Esophageal Cancer Radiotherapy

ObjectiveTo evaluate the feasibility and geometric accuracy of different thoracicanatomical structures used as landmarks for image registration in upper andmid-thoracic esophageal cancer radiation therapy.MethodsThirty-five patients with upper/mid-thoracic esophageal cancer were included inthis study. CT-on-rail was used for patient position verification before eachradiotherapy treatment. The “spine”, and the other five thoracic anatomical structures,including “aorta”,“carina”,“manubrium sterni”,“clavicle”,“apex pulmonis” wereselected as landmarks to analyze the geometric accuracy in the directions ofleft-right(LR), superoinferior(SI), and anteroposterior(AP), respectively.Results1．The position verification were conducted in all of the35esophageal cancerpatients during the first week of radiotherapy treatment, and the position verificationwere conducted every other day after that. In total,662pieces of verification films were collected.2．When the “spine” was selected as landmark for image registration, the setuperrors in the directions of LR, SI and AP were0.31±0.38cm,0.36±0.45cm and0.22±0.26cm, respectively. If the “aorta” was used as landmark, the setup errors in thethree directions were0.32±0.39cm,0.52±0.44cm and0.21±0.26cm, respectively.When the “carina” was selected as landmark for image registration, the setup errors inthe directions of LR, SI and AP were0.27±0.37cm,0.56±0.56cm, and0.32±0.30cm,respectively. If the “manubrium sterni” was used as landmark, the setup errors in thethree directions were0.31±0.32cm,0.60±0.44cm and0.29±0.25cm, respectively.When the “clavicle” was selected as landmark for image registration, the setup errorsin the directions of LR, SI and AP were0.30±0.27cm,0.55±0.38cm and0.23±0.24cm,respectively. If the “apex pulmonis” was used as landmark, the setup errors in thethree directions were0.32±0.41cm,0.44±0.63cm, and0.25±0.28cm, respectively.3．Large-scale positioning movements were noted7times in the662pieces ofverification films, the probability was only1.1%; however, the treatment accuracywould be affected if the movements were not adjusted.4．In the direction of LR, there were no statistical differences among the setuperrors that were got from the five landmarks and from “spine”(p＞0.05). In thedirection of SI, the setup errors which was got using “clavicle” and “apex pulmonis”were significantly larger than those using “spine” as landmark for image registration,p=0.048and0.037, respectively. In the direction of AP, compared with the setuperrors using “spine” as landmark, the setup errors were much larger when “carina”and “manubrium sterni” used as landmarks, p=0.000and0.036, respectively.5．There were no statistical correlations between setup errors and gender, age,disease stage.ConclusionFor the patients with upper/mid-thoracic esophageal cancer who receivedradiation therapy, we don’t suggest “clavicle” and “apex pulmonis” be served as landmarks for image registration in the direction of SI, neither do we recommend the“carina” and “manubrium sterni” be selected as landmarks in the direction of AP.Mobility was limited in the LR direction for all of the thoracic anatomical structures.However, in the direction of SI, only “spine”,“aorta”,“carina”,“manubrium sterni”can be selected as landmarks; in the direction of AP, only “spine”,“aorta”,“clavicle”,and “apex pulmonis” can be used as landmarks for image registration.