Skip to main content

Impact of metal artifact reduction algorithm on gross tumor volume delineation in tonsillar cancer: reducing the interobserver variation

Abstract

Background

Patients with tonsillar cancer (TC) often have dental fillings that can significantly degrade the quality of computed tomography (CT) simulator images due to metal artifacts. We evaluated whether the use of the metal artifact reduction (MAR) algorithm reduced the interobserver variation in delineating gross tumor volume (GTV) of TC.

Methods

Eighteen patients with TC with dental fillings were enrolled in this study. Contrast-enhanced CT simulator images were reconstructed using the conventional (CTCONV) and MAR algorithm (CTMAR). Four board-certified radiation oncologists delineated the GTV of primary tumors using routine clinical data first on CTCONV image datasets (GTVCONV), followed by CTCONV and CTMAR fused image datasets (GTVMAR) at least 2 weeks apart. Intermodality differences in GTV values and Dice similarity coefficient (DSC) were compared using Wilcoxon’s signed-rank test.

Results

GTVMAR was significantly smaller than GTVCONV for three observers. The other observer showed no significant difference between GTVCONV and GTVMAR values. For all four observers, the mean GTVCONV and GTVMAR values were 14.0 (standard deviation [SD]: 7.4) cm3 and 12.1 (SD: 6.4) cm3, respectively, with the latter significantly lower than the former (p < 0.001). The mean DSC of GTVCONV and GTVMAR was 0.74 (SD: 0.10) and 0.77 (SD: 0.10), respectively, with the latter significantly higher than that of the former (p < 0.001).

Conclusions

The use of the MAR algorithm led to the delineation of smaller GTVs and reduced interobserver variations in delineating GTV of the primary tumors in patients with TC.

Peer Review reports

Background

Radiotherapy (RT) with or without chemotherapy is an organ preservation therapy commonly indicated for tonsillar cancer (TC) [1]. High-precision RT techniques such as intensity-modulated radiotherapy and volumetric-modulated arc therapy have become increasingly used for the treatment of TC [2]. As these RT techniques are characterized by highly conformal dose distributions, the treatment success highly depends on the accurate definition of gross tumor volume (GTV). Delineating GTV based on computed tomography (CT) simulator images is required for the RT planning process for TC. However, patients with TC often have dental fillings with metal artifacts in CT scans. Metal artifacts can significantly degrade the quality of simulation CT images, obscure visualization of the primary tumor, and therefore result in a large degree of interobserver variations in delineating GTV of primary tumors [3].

Recently, metal artifact reduction (MAR) algorithms have been increasingly used for CT imaging for the diagnosis and RT planning of head and neck cancer [4,5,6,7]. The commercially available software Smart MAR (GE Healthcare, Chicago, IL, USA) reduces photon starvation, beam hardening, and streak artifacts caused by high z metals in the body [5]. However, the usefulness of MAR algorithms in delineating the GTV of patients with TC is not fully discussed. Therefore, this study aimed to evaluate whether the use of MAR algorithms reduces interobserver variations in delineating the GTV of TC.

Methods

Patients

This retrospective study was approved by the institutional review board of our hospital. Between July 2019 and August 2021, 21 patients with pathologically confirmed tonsillar squamous cell carcinoma with dental fillings underwent pretreatment contrast-enhanced MR imaging within 4 weeks and contrast-enhanced [18F]-fluoro-2-deoxy-D-glucose (FDG)–positron emission tomography (PET)/CT within 6 weeks before RT planning contrast-enhanced CT imaging in our hospital [2]. Three patients were excluded due to the presence of superficial T1 lesions [3]. Finally, the study population consisted of 18 patients, comprising 12 males and 6 females (median age 57 [range 48–88] years). Patients were categorized according to the Union for International Cancer Control TNM staging system, 8th edition: 13 patients with T2, 4 with T3, and 1 with T4 tumors. The p16 status was positive in 15 and negative in 3 patients.

Imaging protocol

The pretreatment imaging protocols were described elsewhere [2, 8]. RT planning CT images with iodinated contrast media were acquired with a 16-row CT scanner (Discovery RT; GE Healthcare, Chicago, IL, USA). During CT scanning, patients were positioned supine with arms by their sides. A pillow and a thermoplastic mask dedicated to RT were used [9, 10]. A total of 80–100 ml of contrast media (Omnipaque 300, Daiichi-Sankyo, Tokyo, Japan) was injected at a 2 ml/s rate using a 22-G intravenous catheter placed in an antecubital vein. The CT scan was conducted 50–60 s after the initiation of contrast injection with helical mode, 120 kVp, Smart mA automatic exposure control (GE Healthcare, Chicago, IL, USA), 2.5-mm slice thickness, and 650-mm field of view (FOV). CT images were reconstructed using 500-mm FOV using the conventional (CTCONV) and Smart MAR algorithms (CTMAR). CTMAR removed metal artifacts based on the following steps: (1) identification of corrupted samples in the projection that corresponds to metallic objects, (2) generating inpainted data by replacing metal corrupted projections with the corrected data generated using the forward projection of the classified image, and (3) generating the final corrected projection by combining the original and inpainted projection data [11]. CTCONV and CTMAR images were transferred to the RT planning system (Pinnacle3 9.10; Philips Medical Systems, Fitchburg, MA, USA). Registration of images was performed with the RT planning system through hardware arrangement.

GTV definition

GTVs of primary tumors were delineated independently by four board-certified radiation oncologists with 5–18 years of experience. Observers were provided with routine clinical data (i.e., contrast-enhanced FDG/PET-CT and contrast-enhanced MR images and endoscopy videos) and asked to contour the GTV of each primary tumor on axial slices of CTCONV datasets. For each case, observers defined the GTV first on CTCONV image datasets (GTVCONV), followed by CTCONV and CTMAR fused image datasets (GTVMAR) at least 2 weeks apart to minimize memory bias and fatigue [9, 12]. When delineating GTVCONV, observers could not view CTMAR images. When delineating CTMAR, observers could view both CTCONV and CTMAR images. Observers could adjust the window level and width on their preferences for delineating GTV.

Evaluation of intermodality differences and observer variations

For each of 18 cases, 8 sets of GTVs were delineated: 4 (observers) × 2 (modalities). Analyses were performed based on three-dimensional volume. GTVCONV and GTVMAR values defined by 4 observers were calculated to evaluate intermodality (CTCONV images vs. CTCONV and CTMAR fused images) differences in GTVs. For geometric interobserver comparisons, the Dice similarity coefficient (DSC), which is used to measure the similarity between two samples, was calculated using the following equation:

$${\text{DSC}} = 2 \times {\text{A}} \cap {\text{B}}/\left( {{\text{A}} + {\text{B}}} \right)$$

where A ∩ B is the volume of the intersection between two GTVs of A and B. The DSC ranged from 0 (no overlap) to 1 (perfect match) [10, 13, 14]. The DSC was calculated as the mean DSC of all possible pair combinations for both GTVCONV and GTVMAR [15].

Statistical analysis

Intermodality differences in GTV values and DSCs were compared using Wilcoxon’s signed-rank test. Statistical calculations were performed using the SPSS software version 25.0 (IBM, Armonk, NY, USA). Differences with p values of < 0.05 were considered statistically significant.

Results

The mean GTVCONV and GTVMAR values for four observers are shown in Table 1. GTVMAR was significantly smaller than GTVCONV for three observers. The other observer showed no significant difference between GTVCONV and GTVMAR values. For all four observers, the mean GTVCONV and GTVMAR values were 14.0 cm3 (standard deviation [SD]: 7.4) cm3 and 12.1 (SD: 6.4) cm3, respectively, indicating that GTVMAR was significantly smaller than GTVCONV (p < 0.001, Table 1).

Table 1 GTVCONV and GTVMAR values for four observers

The mean DSCs of GTVCONV and GTVMAR were 0.74 (SD: 0.10) and 0.77 (SD: 0.10), respectively, indicating that the DSC of GTVMAR was significantly higher than that of GTVCONV (p < 0.001, Table 2). Figure 1 shows a representative patient with DSC of GTVMAR higher than that of GTVCONV.

Table 2 Dice similarity coefficient of GTVCONV and GTVMAR for each pair and all four observers
Fig. 1
figure 1

Examples of gross tumor volume (GTV) in patients with tonsillar cancer. Each panel includes axial (left), sagittal (center), and coronal (right) images. A GTVs were delineated by four observers using conventional computed tomography (CT) images. The mean Dice similarity coefficient (DSC) was 0.80 with a mean GTV value of 23.7 cm3. B GTVs were delineated by four observers based on conventional and metal artifact reduction fused CT images. The mean DSC improved to 0.83 with a mean GTV value of 21.6 cm3 based on the combination of conventional and metal artifact reduction CT images

Discussion

RT for TC is associated with acute and late toxicities, including mucositis, dermatitis, taste dysfunction, xerostomia, and osteoradionecrosis. Therefore, an inappropriately large definition of target volumes may lead to deterioration in the quality of life of patients. Our study suggested that the addition of CTMAR to CTCONV images delineated smaller GTVs than CTCONV images alone in patients with TC. One possible reason is that observers unnecessarily included invisible areas due to metal artifacts in GTVCONV images to prevent marginal miss. Abelson et al. evaluated the effects of using the MAR technique on GTV delineation in 8 patients with TC [3]. Two radiation oncologists independently delineated the GTV of the primary tumor for each patient based on non-MAR CT (GTVnonMDT) and MAR CT (GTVMDT) images. GTVnonMDT and GTVMDT values of axial slices with metal artifacts were not significantly different. However, the number of patients was extremely low and may have thus yielded the difference between our results and theirs. The introduction of CTMAR images into RT planning may prevent unnecessary toxicities in patients with TC by reducing target volumes. Conversely, there is a possibility that smaller GTVs yielded using CTMAR images result in inappropriate dose distribution with unintended underdosing to the actual target volume. These benefits and risks should be considered when introducing CTMAR images into RT planning.

Previous studies evaluated the ability of MAR algorithms to improve organ contouring in RT planning. Kohan et al. used CT images of 11 patients with metal artifacts in the head and neck regions [16]. Five independent observers with 0–6 years of experience including a medical student performed area measurements of selected normal organ structures, such as masseter muscles and tongues on non-MAR CT and MAR CT image slices with metallic objects and non-MAR CT slices without metallic objects as control. The intraclass correlation coefficient (ICC) was calculated to assess interobserver variations. For all observers, ICCs for non-MAR CT, MAR CT, and control non-MAR CT image slices were 0.903, 0.948, and 0.985 with outliers and 0.884, 0.971, and 0.989 without outliers, respectively. For experienced observers, ICCs for non-MAR CT, MAR CT, and control non-MAR CT image slices were 0.904, 0.979, and 0.976 with outliers and 0.934, 0.975, and 0.990 without outliers, respectively. They suggested the use of MAR algorithms greatly reduced the interobserver variation. Our study results suggested that the addition of CTMAR to CTCONV images reduced interobserver variations compared with CTCONV images alone in delineating GTV of primary tumors in patients with TC. Hansen et al. evaluated whether the introduction of MAR algorithms reduces interobserver variations in delineating GTV based on CT images of 11 patients with oropharyngeal cancer (OPC) [13]. Three experienced radiation oncologists and one experienced radiologist independently delineated the GTV of primary tumors (GTV-T) based on non-MAR and MAR CT images. The mean DSCs of GTV-T for non-MAR CT and MAR CT images were 0.60 (SD: 0.24) and 0.61 (SD: 0.20), respectively, showing no difference between the two modalities. The possible reason for the difference between our results and theirs are as follows: (1) the number of their patients was too small and (2) they included patients with OPC other than TC. Greater consistency in delineating GTV with MAR CT images should reduce the influence of potential variability during the RT planning process [17].

Our study has some limitations. First, this was a retrospective study with a relatively small number of patients, although it was larger than previous studies. Second, no pathologic gold standard has been established, which is unavoidable in this type of study. Third, the effects of adding MAR CT images based on dose distributions or treatment outcomes were not evaluated. Further prospective trials based on RT planning simulation using MAR CT images will be required to confirm the clinical benefits for patients with TC.

Conclusions

The use of MAR CT images in addition to conventional CT images led to the delineation of smaller GTVs and reduced interobserver variations in delineating GTV of primary tumors in patients with TC.

Availability of data and materials

The data that support the study findings are available from the corresponding author; however, restrictions apply to the data availability, which was used under license for the present study, and thereby are not publicly available. Data are, however, available from the authors upon reasonable request and with permission of the institutional research ethics board of Kumamoto University Hospital.

Abbreviations

RT:

Radiotherapy

TC:

Tonsillar cancer

GTV:

Gross tumor volume

CT:

Computed tomography

MAR:

Metal artifact reduction

FDG:

[18F]-fluoro-2-deoxy-D-glucose

PET:

Positron emission tomography

FOV:

Field of view

DSC:

Dice similarity coefficient

ICC:

Intraclass correlation coefficient

References

  1. National Comprehensive Cancer Network: NCCN Clinical Practice Guidelines in Oncology, Head and Neck Cancers (Version 2. 2022).

  2. Toya R, Saito T, Fukugawa Y, Matsuyama T, Matsumoto T, Shiraishi S, et al. Prevalence and risk factors of retro-styloid lymph node metastasis in oropharyngeal carcinoma. Ann Med. 2022;54(1):436–41.

    Article  CAS  Google Scholar 

  3. Abelson JA, Murphy JD, Wiegner EA, Abelson D, Sandman DN, Boas FE, et al. Evaluation of a metal artifact reduction technique in tonsillar cancer delineation. Pract Radiat Oncol. 2012;2(1):27–34.

    Article  Google Scholar 

  4. Hirata K, Utsunomiya D, Oda S, Kidoh M, Funama Y, Yuki H, et al. Added value of a single-energy projection-based metal-artifact reduction algorithm for the computed tomography evaluation of oral cavity cancers. Jpn J Radiol. 2015;33(10):650–6.

    Article  Google Scholar 

  5. Katsura M, Sato J, Akahane M, Kunimatsu A, Abe O. Current and novel techniques for metal artifact reduction at CT: practical guide for radiologists. Radiographics. 2018;38(2):450–61.

    Article  Google Scholar 

  6. Andersson KM, Dahlgren CV, Reizenstein J, Cao Y, Ahnesjo A, Thunberg P. Evaluation of two commercial CT metal artifact reduction algorithms for use in proton radiotherapy treatment planning in the head and neck area. Med Phys. 2018;45(10):4329–44.

    Article  CAS  Google Scholar 

  7. Puvanasunthararajah S, Fontanarosa D, Wille ML, Camps SM. The application of metal artifact reduction methods on computed tomography scans for radiotherapy applications: a literature review. J Appl Clin Med Phys. 2021;22(6):198–223.

    Article  Google Scholar 

  8. Toya R, Saito T, Matsuyama T, Kai Y, Shiraishi S, Murakami D, et al. Diagnostic value of FDG-PET/CT for the identification of extranodal extension in patients with head and neck squamous cell carcinoma. Anticancer Res. 2020;40(4):2073–7.

    Article  CAS  Google Scholar 

  9. Toya R, Matsuyama T, Saito T, Imuta M, Shiraishi S, Fukugawa Y, et al. Impact of hybrid FDG-PET/CT on gross tumor volume definition of cervical esophageal cancer: reducing interobserver variation. J Radiat Res. 2019;60(3):348–52.

    Article  CAS  Google Scholar 

  10. Kai Y, Arimura H, Toya R, Saito T, Matsuyama T, Fukugawa Y, et al. Comparison of rigid and deformable image registration for nasopharyngeal carcinoma radiotherapy planning with diagnostic position PET/CT. Jpn J Radiol. 2020;38(3):256–64.

    Article  CAS  Google Scholar 

  11. Pal D, Dong S, Genitsarios I, Hsieh J. Smart Metal Artifact Reduction (MAR). General Electric Healthcare Company. 2013.

  12. Breen SL, Publicover J, De Silva S, Pond G, Brock K, O’Sullivan B, et al. Intraobserver and interobserver variability in GTV delineation on FDG-PET-CT images of head and neck cancers. Int J Radiat Oncol Biol Phys. 2007;8(3):763–70.

    Article  Google Scholar 

  13. Hansen CR, Christiansen RL, Lorenzen EL, Bertelsen AS, Asmussen JT, Gyldenkerne N, et al. Contouring and dose calculation in head and neck cancer radiotherapy after reduction of metal artifacts in CT images. Acta Oncol. 2017;56(6):874–8.

    Article  Google Scholar 

  14. Shimohigashi Y, Doi Y, Kouno Y, Yotsuji Y, Maruyama M, Kai Y, et al. Image quality evaluation of in-treatment four-dimensional cone-beam computed tomography in volumetric-modulated arc therapy for stereotactic body radiation therapy. Phys Med. 2019;68:10–6.

    Article  Google Scholar 

  15. Hagen M, Kretschmer M, Wurschmidt F, Gauer T, Giro C, Karsten E, et al. Clinical relevance of metal artefact reduction in computed tomography (iMAR) in the pelvic and head and neck region: multi-institutional contouring study of gross tumour volumes and organs at risk on clinical cases. J Med Imaging Radiat Oncol. 2019;63(6):842–51.

    Article  Google Scholar 

  16. Kohan AA, Rubbert C, Vercher-Conejero JL, Partovi S, Sher A, Kolthammer JA, et al. The impact of orthopedic metal artifact reduction software on interreader variability when delineating areas of interest in the head and neck. Pract Radiat Oncol. 2015;5(4):e309–15.

    Article  Google Scholar 

  17. Rasch C, Steenbakkers R, van Herk M. Target definition in prostate, head, and neck. Semin Radiat Oncol. 2005;15(3):136–45.

    Article  Google Scholar 

Download references

Acknowledgements

Not applicable.

Funding

None.

Author information

Authors and Affiliations

Authors

Contributions

YF developed the study design; collected and interpreted data; and drafted the manuscript. RT developed the study design and imaging protocol; collected, analyzed, and interpreted the data; and wrote the manuscript. T. Matsuyama and TW collected and interpreted data. YS and YK developed the imaging protocol, collected the data, and performed the statistical analysis. T. Matsumoto and NO developed the study design and interpreted the data. All authors have read and approved the final manuscript.

Corresponding author

Correspondence to Ryo Toya.

Ethics declarations

Ethical approval and consent to participate

This study received full approval from the institutional research ethics board of Kumamoto University Hospital (No. 2281) and it conformed to the principles of the Helsinki Declaration. The requirement of individual participant consent was waived by the research ethics board of Kumamoto University Hospital.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Fukugawa, Y., Toya, R., Matsuyama, T. et al. Impact of metal artifact reduction algorithm on gross tumor volume delineation in tonsillar cancer: reducing the interobserver variation. BMC Med Imaging 22, 161 (2022). https://doi.org/10.1186/s12880-022-00889-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12880-022-00889-0

Keywords

  • Metal artifact reduction
  • Radiotherapy
  • Head and neck cancer
  • Oropharyngeal carcinoma
  • Radiotherapy planning
  • Gross tumor volume