FDG PET/CT interobserver agreement in head and neck cancer: FDG and CT measurements of the primary tumor site.

BACKGROUND: Interobserver agreement in measuring positron emission tomography (PET)/computed tomography (CT) parameters is critical for head and neck cancer management. The purpose of this study is to assess the interobserver agreement when measuring the standardized uptake value (SUV) parameters and the diameter of primary head and neck tumors between readers with varied experience in PET/CT. METHODS: PET/CT images of 47 patients (59.8 ± 10.6 years, range 26-86 years) with head and neck cancer who had a baseline PET/CT at our institution (January 2007-May 2009) were retrieved and independently reviewed by four readers with varying experience reading PET/CT. Novice reader 1 was a preinternship medical student; novice reader 2 was a fourth-year radiology resident; expert reader 1 was a board-certified radiologist with a nuclear radiology fellowship and a junior faculty; and expert reader 2 was a dual radiology and nuclear medicine board-certified radiologist and a senior faculty. Intraclass correlation coefficients (ICCs) were estimated separately across novice and expert readers, between novice 1 and expert 1, and between novice 2 and expert 1 readers. RESULTS: The ICCs (95% confidence interval) for primary tumor diameter, primary tumor SUVmax, and primary tumor SUVmax normalized to liver SUVmean within novices were 0.36 (0.09-0.59) (slight agreement), 0.60 (0.38-0.80) (fair agreement), and 0.91 (0.84-0.94) (almost perfect agreement), respectively. ICCs within experts were 0.84 (0.73-0.90) (almost perfect agreement), 0.91 (0.84-0.94) (almost perfect agreement), and 0.91 (0.84-0.95) (almost perfect agreement), respectively. The ICCs between novice 1 and experts are lower than those between novice 2 and experts for all parameters, and this disparity is greater for anatomic diameter than for fluorodeoxyglucose metabolic parameters. CONCLUSION: Fluorodeoxyglucose metabolic parameters have higher interobserver agreement than anatomic diameter measurement and are more robust in the setting of varied reading experiences.

Duke Scholars

Author Rathan Markandan Subramaniam Radiology