A population-based atlas and clinical target volume for the head-and-neck lymph nodes

To develop a population-based three-dimensional lymph node target volume of the head and neck.

The T2 weighted axial magnetic resonance imaging (MRI) images of 35 patients with known head and neck cancer were reviewed. An experienced head and neck radiologist marked each lymph node (LN) electronically. The images were distributed to one of 12 axial levels of the head and neck with each level representing a distinct portion of the neck based on external contours and the presence of anatomic structures. The LNs were marked with five different symbols to distinguish the superior/inferior extent of each LN within each level. With the categorization of each image into a different superior/inferior level of the neck, the registration of the images was limited to two dimensions. Nonlinear transformation accounted for inter-patient differences although no local warping was used. The co-registration used recognizable anatomic landmarks (vertebral body, mandible, maxilla, clivus as well as the sternocleidomastoid muscle, external skin contour, spinal cord) to match the patient anatomy.

In total, 503 images were co-registered with the baseline images. The majority of the co-registrations were of good quality; 361, 122, and 20 image co-registrations scored as global, limited, and poor co-registrations respectively. One thousand and fifty seven LNs were marked, with 122 LNs marked as submandibular and submental LN. Among the levels A, B, C, D, E, F, G, H, 22, 44, 206, 199, 196, 175, 63, 35 LN were marked respectively. Noteworthy anatomic variation was observed among the different nodal groups that are summarized in the representative baseline images.

Image registration of a series of head and neck images generates a valuable population-based lymph node map that can be used to guide the three-dimensional delineation of the elective lymph node target volume. Significant variation in the lymph node location was seen in all LN groups. The medial border of the internal jugular vein can be used as an important landmark structure in delineating the jugular LN clinical target volume and, for that reason, intravenous contrast is recommended to improve visualization. The location of the submandibular LN appear to be limited to the space anterior and lateral to the submandibular gland and are found mostly along the inferior edge of the mandible. The location of the retropharyngeal LN does vary but their location does not appear to vary with any other recognizable axial structure. The lymph node map provides another collaborating piece of evidence in defining the head and neck LN clinical target volume.