Impact of Image Coregistration With Computed Tomography (CT), Magnetic Resonance (MR) and Positron Emission Tomography With Fluorodeoxyglucose (FDG-PET) on Delineation of GTV's in Oropharyngeal, Laryngeal and Hypopharyngeal Tumors

Reviewer: Roberto Santiago, MD
Last Modified: October 8, 2002

Presenter: Jean-Francois Daisne
Presenter's Affiliation: Head and Neck Onc Prog, Univ Catholique de Louvain, St-Luc Univ Hosp, Brussels, Belgium
Type of Session: Scientific


    The delineation of target volumes for three-dimensional radiotherapy that closely conforms to the target tissues has been traditionally based on computed tomography (CT). Although CT has many advantages (wide availability, excellent visualization of bony structures, electron density information, relatively low movement artifact and distortions), newer imaging modalities appear to complement the CT limitations (soft tissue visualization, multiplanar capabilities, contrast). This study evaluates the potential added-value of MR and FDG-PET over CT for the delineation of primary tumor GTV in oropharyngeal, laryngeal and hypopharyngeal tumors.

Materials and Methods

    Twenty nine patients with locally advanced oropharyngeal (n = 10) and laryngeal/hypopharyngeal (n=19) tumors treated with definitive radiotherapy were included in the study. Nine of the 19 patients with laryngeal/hypopharyngeal tumors underwent subsequent en-block salvage resections and their surgical specimens were examined both radiologically and pathologically. CT, MR and FDG-PET examinations were performed before treatment employing patient immobilization (thermoplastic mask) similar to that used for therapy. MR and FDG-PET acquisitions were coregistered to the CT images using an in house developed software allowing a rigid, manual, segmentation-based coregistration. The accuracy of the whole procedure has been measured in the range of 1?2 mm (+/- 2 standard deviations). For CT and MR, GTVs were delineated on axial sections. For PET, GTVs were automatically delineated based on levels of isoactivity, and corrected for air cavities when judged necessary. Differences in individual (GTV-CT, GTV-MR and GTV-PET) and combined (GTV-CT+MR, GTV-CT+PET and GTV-MR+PET) volumes were assessed using a Wilcoxon signed-rank test on coregistered images.


    No significant difference was observed between the average size of the GTV as defined by CT (32.0 cc) and as defined by MR (29.9 cc). However, the GTV as defined using a combination of CT and MR was 31% larger than the GTV as defined by CT alone. This indicates that individual volumes were not totally overlapping. Qualitative analysis indicated that MR was much better in discriminating muscle and parapharyngeal space infiltration. It was also found to be very helpful to delineate regions which were highly contaminated by teeth filling on CT slices. The average GTV as defined by PET (20.3 cc) was smaller than the corresponding volume as defined by CT (p=0.013) or MR (p=0.050). The use of PET only enlarged the CT-defined GTV by 10% and the MR-defined GTV by 18%. When all 3 imaging modalities were combined, the average resulting GTV was 37% larger than the average GTV-CT. When the pathology specimen data was analyzed, the GTV as defined by PET slightly overestimated the gross tumor extent but missed some areas of microscopic extension.

Author's Conclusions

    Definition of GTV in tumors of the head and neck differed substantially depending on the imaging modalities used. The GTV as defined by PET was always smaller than the corresponding volume defined by the other modalities. Individual volumes did not only vary in size but also did not totally overlapped. The comparison of GTV as delineated on macroscopic pathological slices (laryngeal tumors only) with the GTV based on preoperative CT, MR and FDG-PET slices suggest that imaging, even PET, overestimates the gross tumor extent but misses some microscopic extensions.

Clinical/Scientific Implications

    This is a very important paper that was also presented at this years European Society of of Therapeutic Radiology and Oncology in the Plenary session. Deliniation of the target volume is incredibly important in radiation planning. This study shows that CT, MRI, and PET imaging all have their deficiencies in defining the target. This has broad implications in the era of IMRT and conformal radiotherapy as the trend has been to decrease the margins around the GTV and CTV. All modalities evaluated in this study missed some of the microscopic volume when the specimens were analyzed postoperatively. This study should serve as a word of caution to all radiation oncologists.

Oncolink's ASTRO Coverage made possible by an unrestricted Educational Grant from Ortho Biotech.