Inter- and intraobserver variation in HR-CTV contouring: intercomparison of transverse and paratransverse image orientation in 3D-MRI assisted cervix cancer brachytherapy.

Abstract

To analyze agreement between target volumes, delineated by two observers on transverse (T) and paratransverse (perpendicular to the long cervical axis - PT) MR images for cervix cancer brachytherapy. In 13 patients, High Risk-CTV (HR-CTV) was outlined by two observers in T and PT MR image plane, respecting the GYN GEC-ESTRO recommendations for 3D-image based cervix cancer brachytherapy [1]. Contouring time was measured. HR-CTV sizes were compared, and conformity index (CI) was assessed. Interobserver variations in contour-extent along eight radial directions were compared between delineation planes. After applying a standard treatment plan, an intercomparison of DVH-parameters V100, D90, and D100 for the HR-CTV was carried out. Contouring time was slightly longer in T than PT orientation. Interplane CI did not differ significantly between observers (0.72 vs. 0.71), nor did the interobserver CI between planes (0.79 vs. 0.78). Variations in contour-extent between different radial directions and interplane deviations in DVH parameters were non-significant. Contouring in PT, as compared to T plane, allows for a "circumferential view of the cervix" and facilitates comprehensive understanding of spatial relations between the applicator and patho-anatomical structures. It is marked by a lower contouring difficulty and leads to a comparable outcome in terms of DVH parameters. Interobserver inconsistencies can be minimized by systematic training and following the published recommendations.