Evaluation and development of a predictive model for ultrasound-guided investigation of neck metastases.

Ultrasound investigations for the correct identification of lymph node metastases depend on the experience and qualifications of the investigator; thus, model that provides better preoperative evaluation is desired. Data from 290 patients with an upcoming neck dissection were analyzed to compare the preoperative ultrasound assessment of neck metastases with the pathologically proven postoperative neck status. In total, 364 data sets with 200 malignant and 164 benign lymph nodes were explored. The minimal and maximal transverse diameters and their ratio were shown to be especially good parameters for sensitivity, whereas the echostructure and the presence of a hilum were good for specificity. A model incorporating the evaluated markers is presented. The model provides better judgement of neck lymph nodes in a more objective manner. Using logistic regression, five parameters were identified to predict metastases.