Adaptive RN in HN cancer
Comparison of 12 deformable registration strategies in adaptive radiation therapy for the treatment of head and neck tumors,☆☆

https://doi.org/10.1016/j.radonc.2008.04.010Get rights and content

Abstract

Background and purpose

Weight loss, tumor shrinkage, and tissue edema induce substantial modification of patient’s anatomy during head and neck (HN) radiotherapy (RT) or chemo-radiotherapy. These modifications may impact on the dose distribution to both target volumes (TVs) and organs at risk (OARs). Adaptive radiotherapy (ART) where patients are re-imaged and re-planned several times during the treatment is a possible strategy to improve treatment delivery. It however requires the use of specific deformable registration (DR) algorithms that requires proper validation on a clinical material.

Materials and methods

Twelve voxel-based DR strategies were compared with a dataset of 5 patients imaged with computed tomography (CT) before and once during RT (on average after a mean dose of 36.8 Gy): level-set (LS), level-set implemented in multi-resolution (LSMR), Demons’ algorithm implemented in multi-resolution (DMR), DMR followed by LS (DMR-LS), fast free-form deformable registration via calculus of variations (F3CV) and F3CV followed by LS (F3CV-LS). The use of an edge-preserving denoising filter called “local M-smoothers” applied to the registered images and combined to all the aforesaid strategies was also tested (fLS, fLSMR, fDMR, fDMR-LS, fF3CV, fF3CV-LS). All these strategies were compared to a rigid registration based on mutual information (MI, fMI). Chronological and anti-chronological registrations were also studied. The various DR strategies were evaluated using a volume-based criterion (i.e. Dice similarity index, DSI) and a voxel-intensity criterion (i.e. correlation coefficient, CC) on a total of 18 different manually contoured volumes.

Results

For the DSI analysis, the best three strategies were DMR, fDMR-LS, and fDMR, with the median values of 0.86, 0.85 and 0.85, respectively; corresponding inter-quartile range (IQR) reached 9.6%, 10% and 10.2%. For the CC analysis, the best three strategies were fDMR-LS, DMR-LS and DMR with the median values of 0.97, 0.96 and 0.94, respectively; corresponding IQR reached 11%, 9% and 15%. Concerning the time-sequence analysis, the anti-chronological registration (all deformable strategies pooled) showed a better median DSI value (0.84 vs 0.83, p < 0.001) and IQR (11.2% vs 12.4%). For CC, the anti-chronological registration (all deformable strategies pooled) had a slightly lower median value (0.91 vs 0.912, p < 0.001) but a better IQR (16.4% vs 21%).

Conclusions

The use of fDMR-LS is a good registration strategy for HN-ART as it is the best compromise in terms of median and IQR for both DSI and CC. Even though less robust in terms of CC, DMR is a good alternative. None of the time-sequence appears superior.

Section snippets

Patients and image acquisition

The sets of images used were acquired in a previously reported clinical study [4]. Briefly, contrast-enhanced CT scans were acquired for 10 patients before the start of the treatment and during concomitant chemo-radiotherapy, after mean prescribed doses of 14, 25, 35 and 45 Gy. For all the acquisitions, patients were immobilized with a customized thermoplastic mask (Sinmed, Reuwijk, The Netherlands) fixed to a flat tabletop. Contrast-enhanced CT was performed on a spiral CT scanner (MX 8000 IDT,

Image registration accuracy

As illustrated in Fig. 3, all DR strategies significantly improved the DSI compared to the rigid-body registration (Kruskal–Wallis Z test, p < 0.05). On the other hand, 4 out of the 6 DR strategies (LS vs fLS, LSMRvs fLSMR, DMR-LS vs fDMR-LS and F3CV-LS vs fF3CV-LS) showed that using an edge-preserving filter before the registration improved the volume criteria. However, such improvement was very mild and not significant (Kruskal–Wallis Z test, p > 0.05). Comparison of the median values of DSI

Discussion

In this study, we attempted to compare different DR strategies within the framework of adaptive radiotherapy for the treatment of patients with Head and Neck tumors. The registrations were performed in “extreme” conditions, i.e. on images acquired after having delivered 36.8 Gy on average, and thus presenting a large variation in external and internal contours. In summary, fDMR-LS showed better and more consistent results than the other strategies; it was the best strategy according to

Conclusions

In this paper we presented a methodological study for comparing 12 DR strategies within the framework of adaptive radiotherapy for head and neck tumors. The comparisons were based on volume matching estimation and voxel-intensity correspondance. We demonstrated that fDMR-LS is the best compromise and is effective in deforming CT images of a same patient at different phases of his treatment course. Even though it is less robust in terms of voxel-intensity correlation, DMR seems to be an

Acknowledgements

The authors wish to gratefully thank Weiguo Lu, Kenneth J. Ruchala and Gustavo H. Olivera from Tomotherapy Inc. for sharing their knowledge, algorithms and software.

References (36)

Cited by (179)

  • Free-to-use DIR solutions in radiotherapy: Benchmark against commercial platforms through a contour-propagation study

    2020, Physica Medica
    Citation Excerpt :

    Recently, the report of AAPM task group 132 has tackled the issue of DIR software validation, recommending the use of quantitative contour and landmark analysis on physical and computational phantoms [7]. A large body of studies has already been published regarding the performances of both commercial and open source algorithms in different environments and clinical settings [1,2,8–15]. To the best of our knowledge, free-to-use DIR algorithms have never been directly compared with commercial ones.

View all citing articles on Scopus

Financial support: This work was supported by a grant from the Fonds National pour la Recherche Scientifique (FNRS) of Belgium (convention # 7.4583.07), by a grant from the Belgian Federation against Cancer (convention #SCIE 2003-23FR), by a grant from the “Cancéropôle du Nord-Ouest (France)”, by a grant from the Région wallonne of Belgium (convention PAINTER) and by the “Fonds J. Maisin” of the Université catholique de Louvain. John A. Lee is a Postdoctoral Researcher with the FNRS. The authors have no financial relationship with the organizations that sponsored the research.

☆☆

Statement: The authors have had full control of all primary data and agree to allow the journal to review their data if requested.

1

These authors have equally contributed to this paper.

View full text