Journal of Cancer Research and Therapeutics Medknow Publications on behalf of the Association of Radiation Oncologists of India (AROI) ISSN: 0973-1482 EISSN: 1998-4138 Vol. 7, Num. 2, 2011, pp. 168-173

Journal of Cancer Research and Therapeutics, Vol. 7, No. 2, April-June, 2011, pp. 168-173

Original Article

A study on the tumor volume computation between different 3D treatment planning systems in radiotherapy

Ramachandran Prabhakar, Goura Kishor Rath, Kunhi Parambath Haresh, Nalliah Manoharan, Macherla Anjaneyulu Laviraj, Molaiyan Rajendran, Pramod Kumar Julka

Department of Radiation Oncology, Institute Rotary Cancer Hospital, All India Institute of Medical Sciences, New Delhi, India
Correspondence Address: Ramachandran Prabhakar, Department of Radiation Oncology, Institute Rotary Cancer Hospital, All India Institute of Medical Sciences, New Delhi - 110 029, India, prabhakar_smr@hotmail.com

Code Number: cr11039

PMID: 21768705
DOI: 10.4103/0973-1482.82917

Abstract

Keywords: CT scanning, delineation, DICOM, treatment planning systems, tumor volume

Introduction

In external beam radiotherapy, dose volume histogram plays a significant role in plan evaluation and treatment approval. The accuracy of the dose volume histogram indirectly depends on the tumor definition and critical structures and also on the accuracy of the calculation algorithms. Several studies have shown that tumor volume plays a vital role in the local control and survival rates of patients treated with radiotherapy. ^{[1],[2],[3],[4]} The computed tumor volume varies with the algorithm used for computing the volume in different planning systems. Studies have shown that there is a significant difference between the volumes computed by the treatment planning systems. ^[5],[6] Currently, Dicom objects are widely used in radiotherapy as a standard protocol for transferring images and RT datasets. We have multiple treatment planning systems in our department and the usual methodology is to contour the structures in one treatment planning system/contouring station and transfer the images to other treatment planning system depending on the workload. Contouring is usually performed in Eclipse (Somavision) or in TomoCon contouring workstation and the Dicom RT structure set with the CT images are transferred to other planning system for treatment planning. An effort has been made in this study, to compute the tumor volume for different treatment sites which represent different shapes of tumor volume and assess the variation among the planning systems and or contouring stations.

Materials and Methods

Sixty patients from different treatment sites were selected for this study. Ten patients were selected in each site which includes brain tumors, retinoblastoma, head and neck cancers (Tonsil, Nasopharynx, Oropharynx), lung cancer, upper abdominal cancers (gall bladder, stomach, liver), anal canal/rectal cancer to study the differences in volume variation with different shapes of target volume. All the patients underwent computed tomography (CT) scanning on Philips ^TM Brilliance Big Bore 16 slice CT scanner with a slice thickness of 3 mm. ^[7] The CT datasets were transferred to the Eclipse ^TM treatment planning system (ver. 6.5) which is used as a reference contouring station for volume comparison. The Eclipse treatment planning system uses shape-based interpolation algorithm for the volume computation. The clinical target volume was delineated for all the sixty patients and the CT and Dicom-RT structure datasets were exported to Adac Pinnacle (ver. 8.0M, ADAC Laboratories, CA, USA), Oncentra (Oncentra MasterPlan, Nucletron ^TM ), Plato Sunrise Lightning (Nucletron ^TM ), Precise Plan (Elekta ^TM ) and Ergo ++ (Elekta ^TM ) treatment planning system. The CT and the structures datasets were also exported to the TOMOCON (Tetramed ^TM , Slovak republic) contouring work station in DICOM format. The clinical target volume (CTV) for all the sites were recomputed in all the above contouring and treatment planning system and compared with the volume computed with the Eclipse planning system. Routinely, planning target volume (PTV) is generated from CTV by adding a margin to account for the setup uncertainties and organ motion as per ICRU 62 guidelines. ^[8] Most of the treatment planning systems have the provision for automatic generation of PTV from CTV. To check the variation in PTV generation from CTV with a different planning system, 5 and 10 mm margins were used in this study. In the case of head and neck cancers, the planning target volume (PTV) was grown by adding a margin of 5 mm to the clinical target volume (CTV). Similarly, in the case of lung tumor, the PTV was grown from the CTV by adding a margin of 1 cm to the CTV. The generated PTV volume for head and neck and lung cancers were calculated in Pinnacle, Oncentra, Precise and Ergo++ treatment planning systems and volume was compared with the reference PTV generated in the Eclipse treatment planning system. Plato and TOMOCON do not have the provision for automatic computation of PTV from CTV. SPSS 10.0 was used for statistical analysis. The paired sample t-test was used to compare all the planning system/contouring station using eclipse as the reference system. Similarly, repeated measures were performed to study any variation among the planning system/contouring station.

Results

[Table - 1], [Table - 2], [Table - 3], [Table - 4], [Table - 5] and [Table - 6] show the computed clinical target volume for brain tumors, head and neck cancers, retinoblastoma, lung, upper abdominal malignancies, and anal/rectal cancers, respectively. The tables also show the percentage differences of CTV computed in different planning systems/contouring stations with respect to the Eclipse treatment planning system. The values in brackets indicate the range. All the tables show relatively decreased tumor volumes with Tomocon, Oncentra and Ergo++ treatment planning systems as compared to the reference Eclipse treatment planning system. Similarly, all the tables showed a statistically significant increase in the computed volume with Pinnacle as compared to eclipse. Precise showed a statistically significant difference for head and neck, retinoblastoma and upper abdominal cancers. [Table - 7] shows the computed PTV grown from CTV by adding a 0.5 cm margin in head and neck cancers. Similarly, [Table - 8] shows the computed PTV generated from CTV by adding 1 cm margin in lung cancers. The multivariate tests indicate that the overall differences in generating lung and head and neck PTV from their respective CTV for the Eclipse, Oncentra, Precise, Pinnacle, and Ergo++ lead to statistically significant differences (P<0.0001). In the case of head and neck cancers, no statistical difference was observed with Precise on comparing with eclipse. Pinnacle and Oncentra showed statistically significant differences from Eclipse followed by Ergo++. For lung cancers, Pinnacle and Ergo++ showed a statistically significant difference on comparing with eclipse. [Table - 9] shows the overall summated tumor volume statistics of the clinical target volume for all the treatment sites with different treatment planning systems/contouring stations. It is clearly evident from the results that there is a statistically significant difference among the volume computed with different planning systems. The multivariate tests indicate that the overall differences in all the sites across the seven system are statistically highly significant (P<0.0001). The paired sample t-test showed that all the systems showed a statistically significant difference with Eclipse. The summated volume for all the 60 patients were 12061.4, 11848.8, 12194.8, 11757.8, 12225.4, 12261.2 and 11903.2 cc for Eclipse, Oncentra, Plato, Tomocon, Precise, Pinnacle, and Ergo++ treatment planning systems respectively. [Figure - 1] shows the comparison of different planning systems/contouring station with respect to Eclipse and it shows that Pinnacle and Plato computes relatively larger volume as compared to Eclipse. It also depicts that Oncentra and Ergo++ computed lesser volume as compared to eclipse.

Discussion

The advancement in the field of computer sciences has led to the development of three-dimensional treatment planning system which has revolutionized the treatment planning in radiotherapy. Currently, a wide variety of different 3D planning system is available that mandates stringent quality assurance before incorporating them for treatment planning in a radiotherapy setup. Delineation of target volume plays a vital role in radiotherapy treatment outcome. In external beam radiotherapy, the dose volume histogram is used as a standard tool for evaluating a treatment plan and also for comparing the competing plans.. The information provided by the dose volume histogram is based on the definition of tumor and critical structures, and also on the accuracy of the calculation algorithms. The target volume definition is influenced by several parameters such as imaging modality, ^[9,10] inter-observer variability, ^{[11],[12],[13]} patient movement/setup errors, ^{[1],[2],[3],[4],[5],[6],[7],[8],[9],[10],[11],[12],[13],[14],[15],[16]} resolution limitation of the scanner ^[17] and contouring system. ^[6]

In this study, we have assessed the variation of tumor volume between different treatment planning systems and also compared the differences in generating PTV from CTV with different planning systems. Our study clearly shows that none of the planning system exhibited the same volume as compared to the reference contouring station (Eclipse) indicating that the way of computing the volume differs with treatment planning systems. The tumor volume computed in all the treatment planning and contouring workstation showed differences in volume computation as compared to the reference planning system. To be more precise, the Pinnacle treatment planning system showed relatively higher volume with CTV and also the generated PTV as compared to the Eclipse treatment planning system. The main difference in volume computation between eclipse and ADAC pinnacle is that the Eclipse treatment planning system uses a shaped-based interpolation algorithm while Pinnacle uses the summation of voxels with edge voxels weighted to 50% for volume computation. A shape-based interpolation method tries to reproduce structures that closely correspond to anatomical structures and it is based on the distance transform ^[18] that linearly interpolates the shape of the structure between image planes and smoothens the structure boundaries. The overall comparison of the volume with different planning systems showed that Pinnacle calculated relatively larger volume followed by Plato as compared to Eclipse, whereas Tomocon, Oncentra, and Ergo++ showed reduced tumor volume. As far as the variation in CTV to PTV volume generation, Pinnacle showed a relatively higher volume as compared to eclipse planning system. This study shows that while performing multicentre clinical trials which comprises of different centers utilizing different treatment planning system, the variation in the target volume with the planning system should be taken into consideration as the dose volume histogram indirectly depends on the volume calculation.

Conclusion

The accuracy of the reproduction of the clinical target volume with different treatment planning system has been analyzed in this study. The study shows that all the treatment planning system showed variation in computing the tumor volume.

References

Copyright 2011 - Journal of Cancer Research and Therapeutics

The following images related to this document are available:

Photo images