Evaluation of a metal artifact reduction technique in tonsillar cancer delineation

Article Outline

• Abstract
• Introduction
• Methods and materials
  • Patient population
  • Imaging protocol
  • Metal artifact reduction technique
  • PET segmentation
  • Gross tumor volume comparison
  • Image quality interpretation
• Results
• Discussion
• References
• Copyright

Abstract 

Purpose

Metal artifacts can degrade computed tomographic (CT) simulation imaging and impair accurate delineation of tumors for radiation treatment planning purposes. We investigated a Digital Imaging and Communications in Medicine-based metal artifact reduction technique in tonsillar cancer delineation.

Methods and Materials

Eight patients with significant artifact and tonsil cancer were evaluated. Each patient had a positron emission tomography (PET)-CT and a contrast-enhanced CT obtained at the same setting during radiotherapy simulation. The CTs were corrected for artifact using the metal deletion technique (MDT). Two radiation oncologists independently delineated primary gross tumor volumes (GTVs) for each patient on native (CT_nonMDT), metal corrected (CT_MDT), and reference standard (CT_PET/nonMDT) imaging, 1 week apart. Mixed effects models were used to determine if differences among GTVs were statistically significant. Two diagnostic radiologists and 2 radiation oncologists independently qualitatively evaluated CTs for each patient. Ratings were on an ordinal scale from –3 to +3, denoting that CT_MDT was markedly, moderately, or slightly worse or better than CT_nonMDT. Scores were compared with a Wilcoxon signed-rank test.

Results

The GTV_PET/nonMDT were significantly smaller than GTV_nonMDT (P = .004) and trended to be smaller than GTV_MDT (P = .084). The GTV_nonMDT and GTV_MDT were not significantly different (P = .93). There was no significant difference in the extent to which GTV_nonMDT or GTV_MDT encompassed GTV_PET/nonMDT (P = .33). In the subjective assessment of image quality, CT_MDT did not significantly outperform CT_nonMDT. In the majority of cases, the observer rated the CT_MDT equivalent to (53%) or slightly superior (41%) to the corresponding CT_nonMDT.

Conclusions

The MTD modified images did not produce GTV_MDT that more closely reproduced GTV_PET/nonMDT than did GTV_nonMDT. Moreover, the MTD modified images were not judged to be significantly superior when compared to the uncorrected images in terms of subjective ability to visualize the tonsilar tumors. This study failed to demonstrate value of the adjunctive use of a CT corrected for artifacts in the tumor delineation process. Artifacts do make tumor delineation challenging, and further investigation of other body sites is warranted.

Back to Article Outline

Introduction 

Radiotherapy with or without chemotherapy is often employed for definitive treatment of patients with tonsillar cancers. Historically, patients were treated with opposed lateral beams, where a relatively homogenous dose distribution mitigated the need for highly accurate primary tumor delineation. Today, patients are treated with highly conformal techniques based on 3-dimensional treatment planning, often utilizing intensity modulated radiotherapy. The steep dose gradients and narrow treatment margins employed with head and neck intensity modulated radiotherapy are acceptable only when treatment planning and delivery occur with a high degree of accuracy.

In many patients with head and neck cancer, dental fillings comprised of metal alloy or amalgam are present. These fillings may cause significant artifacts on diagnostic computed tomography (CT) and magnetic resonance imaging (MRI).¹, ² In our clinical experience, dental artifacts can significantly degrade simulation imaging, obscure visualization of the primary tumor, and therefore impair accurate delineation of tumors for planning purposes. There may be significant clinical implications caused by poor target delineation, inadequate target coverage, and insufficient normal tissue sparing, including poor tumor control and adverse events.

A variety of techniques have been devised to reduce metal artifact on CT but most require access to raw data from the CT scanner, which often is unavailable, limiting their practicality. A number of techniques utilize the Digital Imaging and Communications in Medicine (DICOM)-formatted images, but none are commercially available for use in radiation treatment planning. A comparison among artifact correction tools suggested the superiority of the metal deletion technique (MDT) in comparison to other techniques.³, ⁴

The objective of this study was to investigate the potential of the DICOM-based MDT tool to improve tonsillar cancer delineation. If effective, this tool could be integrated into the routine radiotherapy planning workflow.

Back to Article Outline

Methods and materials 

Patient population 

After obtaining Institutional Review Board approval, we retrospectively reviewed the charts of patients diagnosed with tonsillar cancer and treated with radiotherapy at our institution from January through December 2009. All patients had a tissue diagnosis of squamous cell carcinoma with pathology review at our institution. Included patients had primary tumors in the tonsillar region, with primary tumor stage of T2 or T3. Patients with T4 tumors were excluded to minimize interobserver bias. No restrictions were made based on nodal stage. Patient CT images were reviewed, and patients were selected if, by subjective judgment, there was significant dental artifact in the region of the primary tumor that could potentially impact target identification. Exclusion criteria included undergoing a pre-radiotherapy surgical procedure more extensive than biopsy or simple tonsillectomy. Patients with recurrent tumor or who underwent prior induction chemotherapy were excluded. Between January 2009 and December 2009, 8 patients with tonsillar cancer treated with radiotherapy met the above criteria and formed the cohort of this study.

Imaging protocol 

The pre-treatment imaging protocol has been previously reported and is briefly described here.⁵ Before undergoing simulation, patients fasted for at least 8 hours. At the time of radiotherapy simulation, patients were positioned supine with arms by their sides. A custom-molded foam cushion (AcuForm, Medtec, Orange City, IA) and a thermoplastic mask (Aquaplast; WFR/Aquaplast Corp, Wyckoff, NJ) were used for head and neck support and immobilization. Patients were injected with 10 to 18 mCi of ¹⁸F-fluorodeoxyglucose, and image acquisition occurred 45 to 60 minutes later. Imaging obtained during simulation included contrast-enhanced CT (CECT) and positron emission tomography (PET). A non-contrast-enhanced CT was also obtained, and used for PET attenuation correction only. All imaging was obtained at the same setting, using the same immobilization device, minimizing systematic registration error. The CECTs were acquired at 0.25-cm intervals. The CT imaging was collected in helical acquisition mode on a GE Discovery ST PET-CT scanner (GE Medical Systems, Milwaukee, WI). Two-dimensional PET imaging was obtained over 3 to 5 minutes of acquisition time per bed position. The 2-dimensional PET data were reconstructed with an ordered set expectation maximization algorithm. Patient setup, radiotracer uptake, and CT and PET acquisition required approximately 90 minutes. All patients had glucose levels between 80 and 120 at the time of PET acquisition.

Metal artifact reduction technique 

Metal artifact reduction was performed using the MDT, an automated technique that works off the baseline DICOM image.³, ⁴ The DICOM image set generated by the scanner has Hounsfield units capped at 3071. No manual bulk-tissue override was done. The images from the scanner were defined as the native uncorrected CECT (CT_nonMDT). Metal artifact reduction began by identifying axial slices with metal, using a cutoff of 3000 Hounsfield units. These slices were forward projected to generate simulated projection data. Projection data that included metal were replaced with values linearly interpolated from adjacent nonmetal data, and filtered backprojection was then used to obtain the initial estimate of the slice without metal. Four iterations of filtered backprojection were then performed. On each iteration, projection data that included metal were replaced with forward projected values from the previous iteration. Finally, metal pixels were copied from the original DICOM file. For those axial slices where no metal was present, the axial slices from the CT_nonMDT were not modified. The resultant image set was defined as the metal deletion technique corrected CECT (CT_MDT).

PET segmentation 

Segmentation of PET data was performed for each patient to generate a relative metabolic tumor volume. The volume in the region of the primary tumor with a standardized uptake value (SUV) greater than or equal to 50% of SUV_max is the metabolic tumor volume (MTV_50%), which we previously demonstrated to be prognostic in head and neck cancer.⁵ We used the MIM Maestro software suite, version 5.1 (MIM Software, Inc, Cleveland, OH) for the PET segmentation.

Gross tumor volume comparison 

Two radiation oncologists (W.H., E.W.) independently delineated the gross tumor volume (GTV) of the primary tumor for each patient. No lymph nodes were contoured. Observers had access to reports of clinical and imaging findings, including description of tumor stage, physical exam findings, and radiologic findings; PET reports were available in all cases, and MRI reports were available in 4 patients. Observers separately contoured the GTV on CT axial slices on the 3 image sets: (1) CT_nonMDT; (2) CT_MDT; and (3) reference standard CT (CT_PET/nonMDT), which included both PET and CT_nonMDT data. For each patient, either CT_MDT or CT_nonMDT was selected randomly for the first contouring session. The remaining CT_MDT or CT_nonMDT was selected for the second contouring session, which occurred 1 week later. Subsequently, observers delineated GTVs using CT_PET/nonMDT an additional week later. For GTV_PET/nonMDT, observers could view the fused PET and CT_nonMDT separately or concurrently as a superimposed pair. The PET was augmented with the MTV_50% contour, which observers were instructed to interpret as the metabolic tumor volume. Besides MTV_50%, no restrictions were made with respect to windowing and leveling. The delineation of GTV_PET/nonMDT, which took into account MTV_50% information, was at the discretion of the observer. Taken together, each observer contoured the GTV for each patient on 3 separate sessions, each separated by at least 1 week, intended to minimize memory bias. Contouring on CT_nonMDT, CT_MDT, and CT_PET/nonMDT produced 3 GTV volumes for comparison: GTV_nonMDT, GTV_MDT, and GTV_PET/nonMDT. Each GTV volume was split into 2 sub-volumes, one including axial slices with metal artifact and the other without metal artifact. The GTV_MDT sub-volume without metal artifact was, in fact, delineated on an image set identical to that for the corresponding GTV_nonMDT sub-volume. Therefore, the GTV_MDT sub-volume in the region without artifact was renamed GTV_nonMDT#2.

Comparisons between respective volumes included a concordance index (CI) and geographic miss index (GMI). The CI was defined as

The GMI was defined as

Baseline variability on CT imaging was obtained by comparing GTV_nonMDT and GTV_nonMDT#2. Potential differences after metal correction were evaluated by comparing GTV_nonMDT versus GTV_MDT in comparison to GTV_PET/nonMDT. Mixed effects models were used to assess for differences between GTVs. The data points were not normally distributed, and therefore a logarithmic transformation was applied. Mixed effects modeling accounted for multiple observers and the fact that CT_nonMDT, CT_MDT, and CT_PET/nonMDT were all correlated. The mixed effects model assumed an unstructured correlation matrix. All analyses were done with SAS version 9.2 (SAS Institute Inc, Cary, NC).

Image quality interpretation 

Two diagnostic radiologists (D.A., D.S.) independently qualitatively evaluated the CTs of each patient. Also, after repeated GTV delineation as above, the 2 radiation oncologists performed qualitative image comparison. Each observer rated the difference between CT_nonMDT and CT_MDT on an ordinal scale ranging from (−3) to +3. A score of (−3), (−2), or (−1) indicated that CT_MDT, when compared with CT_nonMDT, was markedly, moderately, or slightly inferior, respectively, whereas a score of 1, 2, or 3 indicated a corresponding scale of relative superiority of CT_MDT, and a score of 0 denoted equivalency. The level of improvement was based on observer interpretation of the diagnostic quality of the images with respect to visualizing the primary tumor. Images were viewed using the MIM Maestro software suite, version 5.1 (MIM Software, Inc, Cleveland, OH). Each observer's scores were compared to the null hypothesis of 0, with a Wilcoxon signed-rank test.

Back to Article Outline

Results 

The bulk of each tonsillar tumor was delineated on axial slices without artifact, although in all cases metal artifact was present in a portion of each contoured GTV. In 6 of 8 patients, the superior extent of GTV_PET/nonMDT was within axial slices with artifact for 1 or both observers. The mean percent of each GTV_PET/nonMDT that fell within artifact regions was 27%. In 3 of 8 patients, no portion of MTV_50% was within the artifact region, and in the remaining 5 patients, an average of 18% of MTV_50% fell within the artifact region (Table 1).

Table 1. Extent of tumor in region of metal artifact

CT, computed tomography; GTV, gross tumor volumes; MDT, metal deletion technique; PET, positron emission tomography; SUV, standardized uptake value.

aMTV50% = volume of tumor with SUV greater than or equal to 50% of maximum SUV.

bGTVPET/nonMDT = gross tumor volume delineated on native uncorrected CT with aid of PET.

In axial slices where no metal artifact was present, GTV_PET/nonMDT sub-volumes were significantly smaller than GTV_nonMDT sub-volumes (P = .020). This finding was reproduced on GTV_nonMDT#2 sub-volumes (P = .009). The mean concordance indexes of GTV_nonMDT and GTV_nonMDT#2 sub-volumes were both 1.4, with standard deviations of 0.4 and 0.3, respectively. In this region, the mean and standard deviation of GTV_nonMDT sub-volumes (12.4, 8.0) were not significantly different (P = .98) than on the GTV_nonMDT#2 sub-volumes (12.6, 6.9). Differences on an individual patient basis between GTV_nonMDT and GTV_nonMDT#2 reflect baseline intraobserver variability in repeat contouring.

On axial slices where metal artifact was present, GTV_PET/nonMDT sub-volumes were also significantly smaller than GTV_nonMDT sub-volumes (P = .004) and trended to smaller than GTV_MDT sub-volumes (P = .084). The mean and standard deviation concordance indexes for GTV_nonMDT (1.9, 0.6) and GTV_MDT (2.2, 1.1) were greater in regions with artifact than without, although this difference was not statistically significant for GTV_nonMDT (P = .13) nor for GTV_MDT (P = .31). The GTV_nonMDT and GTV_MDT sub-volumes were not significantly different in regions with artifact (P = .93) (Table 2).

Table 2. Concordance index

CT, computed tomography; GTV, gross tumor volumes; MDT, metal deletion technique.

aGTVnonMDT = gross tumor volume delineated on native uncorrected CT.

bGTVMDT = gross tumor volume delineated on metal deletion technique-corrected CT.

Although GTV_PET/nonMDT were smaller than GTV_nonMDT and GTV_MDT, parts of GTV_PET/nonMDT were not covered by CT-only contours. Lack of target coverage, defined as GMI, was present both in regions with and without artifact. In regions without artifact there was no significant difference in GMI between GTV_nonMDT and GTV_nonMDT#2 sub-volumes, in terms of both volume (P = 0.67) and percent miss (P = .66). Although not significant, the differences in GMI mean and standard deviations between GTV_nonMDT (0.4, 0.5) and GTV_nonMDT#2 (0.8, 0.7) show a degree of baseline observer variability in volumetric contouring (Table 3).

Table 3. Geographic miss index

CT, computed tomography; GTV, gross tumor volumes; MDT, metal deletion technique.

aGTVnonMDT = gross tumor volume delineated on native uncorrected CT.

bGTVMDT = gross tumor volume delineated on metal deletion technique corrected CT.

In regions with artifact, on an individual patient basis, there were cases in which coverage by GTV_MDT was superior to GTV_nonMDT, and vice-versa. In aggregate, the GMI differences fell within the baseline observer variability described in regions without artifact. The GMI mean and standard deviation for GTV_nonMDT (0.2, 0.2) and GTV_MDT (0.2, 0.4) were very similar. In fact, there was no significant difference in GMI between GTV_nonMDT and GTV_MDT sub-volumes in regions with artifact, in terms of both volume (P = .33) and percent miss (P = .47) (Table 3).

In the subjective assessment of image quality, CT_MDT did not significantly outperform CT_nonMDT. In the majority of cases, the observer rated the CT_MDT equivalent to (53%) or slightly superior (41%) to the corresponding CT_nonMDT. In 1 of 32 comparisons CT_MDT was considered slightly inferior, and in 1 of 32 comparisons CT_MDT was judged moderately better. In no cases was CT_MDT rated markedly better. For each radiation oncologist and diagnostic radiologist, the evaluation of image quality was not statistically significantly different between uncorrected and corrected images (Table 4; Fig 1).

Table 4. Subjective assessment of image quality for visualization of tonsillar tumors

CT, computed tomographic.

Note: Observations of difference made on an ordinal scale. Range (−3,−2,−1, 0,+1,+2,+3) denotes that the quality of the metal deletion technique corrected CT image was markedly, moderately, or slightly worse (−) or better (+) than the native uncorrected CT image.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  Examples of 3 patients with tonsillar cancer, before and after artifact correction with the metal deletion technique (MDT). For each case, corresponding axial images are presented from native uncorrected computed tomographic (CT) and MDT corrected CT.

Back to Article Outline

Discussion 

The GTV_MDT did not more closely reproduce GTV_PET/nonMDT than did GTV_nonMDT, although in select cases GTVs benefited from artifact deletion. Moreover, the MTD modified images were not judged to be significantly superior when compared to the uncorrected images in terms of subjective ability to visualize the tonsilar tumors. Although an artifact reduction technique could improve target delineation, the MDT tool did not improve GTV reproducibility in this highly selective patient cohort, and the appropriate application of artifact correction tools is not yet clear.

We found that tonsillar cancer GTVs based on CT data produced volumes significantly larger than when metabolic imaging was incorporated into the delineation process. This lack of precision with volumetric imaging alone was amplified in regions where artifacts were present. These larger GTV volumes led to nearly adequate, though not complete, coverage of the reference standard tumor volume. But these larger volumes were inappropriate in the context of highly conformal techniques and the need for normal tissue sparing, supporting the need for the inclusion of metabolic imaging, especially when artifacts are present.

There were a number of limitations to this investigation, including the inherent challenges of contouring head and neck targets, where interobserver and even intraobserver variability on repeated contouring can be significant when CT alone is used. We compared contours based on CT-only and PET-enhanced images, although we recognize the weaknesses of PET imaging as a reference standard, including possible errors in registration and limited spatial resolution. Another problem with metabolic imaging is that dental artifact in the non-contrast CT data used for attenuation correction may artificially alter the ¹⁸F-fluorodeoxyglucose uptake values, thereby potentially introducing error in target delineation when PET data are used for treatment planning.⁶ Accounting for errors in attenuation correction is an area of investigation, and is potentially an additional aspect where artifact corrected imaging could improve diagnostic imaging interpretation and simulation imaging for treatment planning.

The power of the study was limited in a number of ways. First, artifact obscured only a small part of each tumor volume. Second, in some cases, the degree of artifact was nearly overwhelming, and even corrected images were still difficult to interpret in the region of the tumor. Third, in the process of contouring in regions with artifact, observers used individualized compensation mechanisms, including interpolation and extrapolation from axial slices without artifact, which to some extent mitigated the artifact problem. Finally, out of necessity, the study was biased against significant results by defining the reference standard using one of the image sets under evaluation, namely the CT_nonMDT; repeating the investigation with a reference standard that instead incorporated the CT_MDT may have led to different results.

The fact that there was no pathologic gold standard, because patients did not undergo surgery as primary management, was an unavoidable limitation of this study. Although no study has clearly demonstrated that PET-enhanced target volumes are clearly more accurate than those delineated by CT data alone, many studies suggest that combining metabolic and anatomic imaging is better than or at least equivalent to anatomic imaging alone in treatment planning.^2,7, ⁸, ⁹, ¹⁰, ¹¹, ¹², ¹³, ¹⁴ However, the optimal method of PET imaging segmentation remains an area of debate,¹⁵, ¹⁶ and metabolic volumes may be highly variable when no clear segmentation method is utilized.¹⁷ The choice of segmentation threshold greatly influences the subsequent metabolic tumor volume,¹⁸, ¹⁹, ²⁰ which may be different from the tumor volume based on anatomic imaging alone.¹⁴, ¹⁹, ²⁰ Fixed-threshold, arbitrary-threshold, adaptive-threshold, and gradient-based techniques exist, none without weaknesses. A number of prior investigations have used 50% of SUV_max as the metabolic threshold.⁵, ⁷, ⁸ Moreover, we noted a link between survival and a fixed-threshold of 50% of SUV_max.⁵ Based on these findings and qualifications, we defined our reference standard as the combination PET and CT_nonMDT, where the metabolic tumor volume was influenced both by the automatically segmented MTV_50% as well as the observer-specific modifications.

All patients with head and neck cancer undergo some form of volumetric imaging, typically contrast-enhanced CT and often gadolinium-enhanced MRI. Dental artifacts may make accurate delineation of the primary tumor on CT very difficult, especially in the oral cavity and adjacent pharynx, and in this setting PET-CT is particularly advantageous.¹, ² For head and neck cancer, contours based on anatomic imaging alone can be highly variable.²¹, ²² The incremental value of MRI in addition to CT is not clear.²³ The majority of studies find a synergistic role for metabolic imaging in combination with volumetric imaging in radiotherapy planning.^2,7, ⁸, ⁹, ¹⁰, ¹¹, ¹², ¹³, ¹⁴ A pattern of failure investigation found that most local recurrences occurred in the metabolically active volume,¹⁰ and 1 study reported improved event-free survival when metabolic imaging was integrated into planning.²⁴

A number of options exist for patients with significant artifacts from high density material. One is the use of megavoltage CT; although soft tissue contrast is diminished, artifacts from high density material could be decreased when compared to conventional kilovoltage CT. Another option is the use of MRI, although it is also subject to image degradation from high-density material. Except for select radiosurgical systems, MRI is not employed as the primary imaging modality in radiotherapy planning and is typically not acquired in the immobilized simulation position, leading to systematic errors when image fusion is inexact. A more straightforward option is a software tool capable of automated artifact reduction that could be integrated into radiotherapy planning workflow.

To our knowledge, this is the first study to evaluate the effect of artifact reduction on target delineation. Even in a highly selective patient cohort, this study failed to demonstrate value in the adjunctive use of a CT corrected for artifacts in the tumor delineation process. Artifacts do make tumor delineation challenging, and further investigation of other body sites is warranted.

Back to Article Outline

References 

1. Baek CH, Chung MK, Son YI, et al. Tumor volume assessment by 18F-FDG PET/CT in patients with oral cavity cancer with dental artifacts on CT or MR images. J Nucl Med. 2008;49:1422–1428
   • View In Article
   • CrossRef
2. Heron DE, Andrade RS, Flickinger J, et al. Hybrid PET-CT simulation for radiation treatment planning in head-and-neck cancers: a brief technical report. Int J Radiat Oncol Biol Phys. 2004;60:1419–1424
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (316 KB)
   • CrossRef
3. Boas FE, Fleischmann D. Evaluation of two iterative techniques for reducing metal artifacts in computed tomography. Radiology. 2011;259:894–902
   • View In Article
   • CrossRef
4. Golden C, Mazin SR, Boas FE, et al. A comparison of four algorithms for metal artifact reduction in CT imaging. [Abstract] In: SPIE Medical Imaging Conference. 2011;
   • View In Article
5. La TH, Filion EJ, Turnbull BB, et al. Metabolic tumor volume predicts for recurrence and death in head-and-neck cancer. Int J Radiat Oncol Biol Phys. 2009;74:1335–1341
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (214 KB)
   • CrossRef
6. Kamel EM, Burger C, Buck A, von Schulthess GK, Goerres GW. Impact of metallic dental implants on CT-based attenuation correction in a combined PET/CT scanner. Eur Radiol. 2003;13:724–728
   • View In Article
   • MEDLINE
7. Koshy M, Paulino AC, Howell R, Schuster D, Halkar R, Davis LW. F-18 FDG PET-CT fusion in radiotherapy treatment planning for head and neck cancer. Head Neck. 2005;27:494–502
   • View In Article
   • MEDLINE
   • CrossRef
8. Paulino AC, Koshy M, Howell R, Schuster D, Davis LW. Comparison of CT- and FDG-PET-defined gross tumor volume in intensity-modulated radiotherapy for head-and-neck cancer. Int J Radiat Oncol Biol Phys. 2005;61:1385–1392
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (1201 KB)
   • CrossRef
9. Geets X, Daisne JF, Tomsej M, Duprez T, Lonneux M, Grégoire V. Impact of the type of imaging modality on target volumes delineation and dose distribution in pharyngo-laryngeal squamous cell carcinoma: comparison between pre- and per-treatment studies. Radiother Oncol. 2006;78:291–297
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (130 KB)
   • CrossRef
10. Soto DE, Kessler ML, Piert M, Eisbruch A. Correlation between pretreatment FDG-PET biological target volume and anatomical location of failure after radiation therapy for head and neck cancers. Radiother Oncol. 2008;89:13–18
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (208 KB)
    • CrossRef
11. Daisne JF, Duprez T, Weynand B, et al. Tumor volume in pharyngolaryngeal squamous cell carcinoma: comparison at CT, MR imaging, and FDG PET and validation with surgical specimen. Radiology. 2004;233:93–100
    • View In Article
    • MEDLINE
    • CrossRef
12. Deantonio L, Beldi D, Gambaro G, et al. FDG-PET/CT imaging for staging and radiotherapy treatment planning of head and neck carcinoma. Radiat Oncol. 2008;3:29
    • View In Article
    • CrossRef
13. Schwartz DL, Ford EC, Rajendran J, et al. FDG-PET/CT-guided intensity modulated head and neck radiotherapy: a pilot investigation. Head Neck. 2005;27:478–487
    • View In Article
    • MEDLINE
    • CrossRef
14. Wang D, Schultz CJ, Jursinic PA, et al. Initial experience of FDG-PET/CT guided IMRT of head-and-neck carcinoma. Int J Radiat Oncol Biol Phys. 2006;65:143–151
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (266 KB)
    • CrossRef
15. Grégoire V, Haustermans K, Geets X, Roels S, Lonneux M. PET-based treatment planning in radiotherapy: a new standard?. J Nucl Med. 2007;48(Suppl 1):68S–77S
    • View In Article
16. Al-Ibraheem A, Buck A, Krause BJ, Scheidhauer K, Schwaiger M. Clinical applications of FDG PET and PET/CT in head and neck cancer. J Oncol. 2009;2009:208725
    • View In Article
17. Riegel AC, Berson AM, Destian S, et al. Variability of gross tumor volume delineation in head-and-neck cancer using CT and PET/CT fusion. Int J Radiat Oncol Biol Phys. 2006;65:726–732
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (262 KB)
    • CrossRef
18. Moule RN, Kayani I, Moinuddin SA, et al. The potential advantages of (18)FDG PET/CT-based target volume delineation in radiotherapy planning of head and neck cancer. Radiother Oncol. 2010;97:189–193
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (385 KB)
    • CrossRef
19. Schinagl DA, Hoffmann AL, Vogel WV, et al. Can FDG-PET assist in radiotherapy target volume definition of metastatic lymph nodes in head-and-neck cancer?. Radiother Oncol. 2009;91:95–100
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (219 KB)
    • CrossRef
20. Schinagl DA, Vogel WV, Hoffmann AL, van Dalen JA, Oyen WJ, Kaanders JH. Comparison of five segmentation tools for 18F-fluoro-deoxy-glucose-positron emission tomography-based target volume definition in head and neck cancer. Int J Radiat Oncol Biol Phys. 2007;69:1282–1289
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (2133 KB)
    • CrossRef
21. Ketting CH, Austin-Seymour M, Kalet I, Unger J, Hummel S, Jacky J. Consistency of three-dimensional planning target volumes across physicians and institutions. Int J Radiat Oncol Biol Phys. 1997;37:445–453
    • View In Article
    • Abstract
    • Full-Text PDF (5333 KB)
    • CrossRef
22. Hermans R, Feron M, Bellon E, Dupont P, Van den Bogaert W, Baert AL. Laryngeal tumor volume measurements determined with CT: a study on intra- and interobserver variability. Int J Radiat Oncol Biol Phys. 1998;40:553–557
    • View In Article
    • Abstract
    • Full-Text PDF (490 KB)
    • CrossRef
23. Geets X, Daisne JF, Arcangeli S, et al. Inter-observer variability in the delineation of pharyngo-laryngeal tumor, parotid glands and cervical spinal cord: comparison between CT-scan and MRI. Radiother Oncol. 2005;77:25–31
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (139 KB)
    • CrossRef
24. Rothschild S, Studer G, Seifert B, et al. PET/CT staging followed by Intensity-Modulated Radiotherapy (IMRT) improves treatment outcome of locally advanced pharyngeal carcinoma: a matched-pair comparison. Radiat Oncol. 2007;2:22
    • View In Article
    • CrossRef