Automatic contouring of brachial plexus using a multi-atlas approach for lung cancer radiation therapy

摘要: Abstract Purpose To demonstrate a multi-atlas segmentation approach to facilitating accurate and consistent delineation of low-contrast brachial plexuses on computed tomographic images for lung cancer radiation therapy. Methods Materials We retrospectively identified 90 patients with treatment volumes near the plexus. Ten representative were selected form an atlas group, their delineated manually. used deformable image registration map each plexus remaining 80 patients. In patient, composite contour was created from 10 individual segmentations using simultaneous truth performance level estimation algorithm. This auto-delineated reviewed modified appropriately patient. also performed leave-one-out tests atlases validate accuracy contouring consistency segmentation. Results The took less than 2 minutes complete. Contour modification 5 compared 20 manual scratch. had mean 3-dimensional (3D) volume overlap 59.2% ± 8.2% 3D surface distance 2.4 mm 0.5 mm. distances between average contours in demonstrated much better contours. auto-segmented did not require substantial modification, by good agreement Dose histograms agreement, showing that editing is clinically acceptable view dosimetric impact. Conclusions Multi-atlas greatly reduced time improved consistency. Editing delineate proved be clinical practice manually