International Journal of Radiation Oncology*Biology*Physics
Physics contributionsFour-dimensional image-based treatment planning: Target volume segmentation and dose calculation in the presence of respiratory motion
Introduction
Variations in organ shape and position can occur on both an interfractional and an intrafractional time scale. Sources of interfractional change include tumor shrinkage or growth, weight changes, and variations in organ filling (e.g., bowel/bladder variations). Sources of intrafractional motion include peristalsis, cardiac motion, and respiration.
These temporal anatomic changes can introduce significant errors in imaging, treatment planning, and treatment delivery. Imprecise knowledge of the target shape and trajectory imposes a strategy of larger field margins, resulting in suboptimal dose conformation. Ideally, accurate organ motion would be included explicitly in treatment planning, on a patient-specific basis to determine its effect on the delivered dose. The focus of this article was to frame the important elements of treatment planning in the presence of respiration-induced organ motion. We analyzed and report our experience concerning the effects of respiratory motion on target delineation, dose calculation, and dose-volume histogram (DVH) analysis using newly available volumetric temporal imaging. The anatomic sites included tumors in the thorax and abdomen.
Treatment planning is frequently based on a single computed tomography (CT) scan, taken during light breathing. Respiratory motion during CT data acquisition can induce severe motion artifacts, resulting in inaccurate assessment of organ shape and location (1, 2, 3, 4). To reduce such artifacts, CT data may be acquired during breath holds (5) or by gating at a specific respiratory phase during imaging (6). Additional data on patient-specific respiratory motion can be obtained by fluoroscopy (7). However, fluoroscopy is limited to planar projections of anatomy. To obtain volumetric three-dimensional (3D) anatomic data during respiration, a time-resolved CT data acquisition protocol is needed.
Once volumetric anatomic maps are acquired, the process of target and normal organ segmentation can begin. Typically, the gross target volume (GTV) is outlined on the treatment-planning scan. This volume is then expanded to a clinical target volume (CTV) to include suspected microscopic spread (8). To ensure sufficient dose coverage of the CTV throughout the treatment course, margins are added to include the geometric uncertainties. Using the new nomenclature of International Commission on Radiation Units and Measurements (ICRU) Report 62 (9), the CTV is expanded to an internal target volume to account for internal target motion; the internal target volume is subsequently expanded to a planning target volume (PTV) to incorporate the daily setup uncertainties.
Geometric margins to account for respiratory motion are usually derived from a combination of experience, direct observation at simulation (e.g., via fluoroscopy), or using values reported in the literature (10). Such margins are tumor site specific, but generally not patient specific. Stevens et al. (11) and van Sörnsen et al. (12) have reported that the tumor location cannot reliably predict the mobility of lung tumors. Although patient-specific target motion can be assessed through two-dimensional projections by fluoroscopy (7), it has its limitations. Studies of patients with abdominal tumors clipped with radiopaque markers have provided data on clip motion only, not explicitly on the regions of the target (or unclipped normal organs). In a retrospective study of treatments of hepatocellular tumors, Rosu et al. (13) reported significant changes in the liver dose if organ motion were included in the dose calculations. Their dose calculations were based on a single breath-hold CT scan and a dose convolution method in which patient-specific motion was obtained by fluoroscopy. To calculate the dose distributions, including temporal anatomic changes in three dimensions, time-resolved CT is needed.
Four-dimensional (4D)-CT data provide the primary image data needed to include explicitly patient-specific respiratory motion into treatment planning to ensure dose coverage of the target throughout the breathing cycle and to calculate the dose distributions for the targets and organs at risk when respiratory motion is present during beam delivery. The accuracy of the 4D-CT scan protocol has been validated through phantom studies and has been reported in detail elsewhere (14). Our initial patient imaging study involved 4D-CT scans of 20 patients, 10 with abdominal tumors and 10 with thoracic tumors. Given these data, issues of data acquisition, assessment of organ motion, dose calculation, and analysis were studied. The explicit inclusion of time-resolved patient data altered the standard workflow of radiotherapy (RT) planning. We report on 4D-CT scanning at Massachusetts General Hospital (MGH) and discuss how the conventional workflow for treatment planning was altered to use 4D-CT.
Section snippets
4D CT scanning
The initial work on 4D-CT has been previously reported (15, 16, 17). The common underlying principle in these approaches has been to acquire axial/helical tomographic images while simultaneously recording the respiratory breathing pattern. These data are then reorganized (resorted) to generate multiple spatiotemporally coherent anatomic data sets.
The 4D-CT scan process used in this study was developed by General Electric Medical Systems (18). CT data were acquired on a General Electric
Motion analysis
Figure 1 shows a coronal slice through the center of the liver for four different CT data sets, including the standard helical data acquisition during light breathing and three 4D-CT data sets at specific respiratory phases of 0%, 20%, and 40%, corresponding to inhale, mid-exhale, and end-exhale. Typical breathing artifacts were present in the standard helical scan. Specifically, the liver edges were jagged or blurred owing to organ motion during data acquisition; consecutive slices under light
Discussion
As shown in Fig. 14, the major tasks in 4D-RT are fundamentally the same as those in 3D-RT currently in practice. The workflow involves the key tasks of image acquisition, target delineation, and treatment planning and delivery. However, the process can be significantly more involved in its most explicit implementation. Image acquisition is marginally more complex. Data acquisition involves monitoring the respiratory cycle during CT scanning, but the total time typically required to take a
Conclusion
We have demonstrated the feasibility of explicitly incorporating patient-specific 4D-CT imaging data in external beam treatment planning. To achieve this, special attention was given to defining the target volumes in the time domain and to apply methods to perform deformable registration during respiration. New technical challenges in performing 4D dose calculations include coping with 10–20 times more imaging data and the associated increase in image segmentation. The output of 4D planning are
Acknowledgments
The authors acknowledge the technical support of General Electric Medical Systems and Varian Medical Systems in the acquisition of four-dimensional CT data; give special thanks to the members of the Department of Radiation Oncology who provided treatment planning and contouring expertise, including Thomas Harris, Judith Adams, and Jonathan Jackson; and acknowledge Andrzej Niemierko for the calculation of EUDs.
References (26)
- et al.
Uncertainties in CT-based radiation therapy planning associated with patient breathing
Int J Radiat Oncol Biol Phys
(1996) - et al.
Impact of respiratory movement on the computed tomographic images of small lung tumors in three-dimensional (3D) radiotherapy
Int J Radiat Oncol Biol Phys
(2000) - et al.
Artifacts in computed tomography scanning of moving objects
Semin Radiat Oncol
(2004) - et al.
Can PET provide the 3D extent of tumor motion for individualized internal target volumes? A phantom study of the limitations of CT and the promise of PET
Int J Radiat Oncol Biol Phys
(2003) - et al.
Improvement of CT-based treatment planning model of abdominal targets using static exhale imaging
Int J Radiat Oncol Biol Phys
(1998) - et al.
Determination of ventilatory liver movement via radiographic evaluation of diaphragm position
Int J Radiat Oncol Biol Phys
(2001) - et al.
Organ motion and its management
Int J Radiat Oncol Biol Phys
(2001) - et al.
Respiratory driven lung tumor motion is independent of tumor size, tumor location, and pulmonary function
Int J Radiat Oncol Biol Phys
(2001) - et al.
Tumor location cannot predict the mobility of lung tumorsA 3D analysis of data generated from multiple CT scans
Int J Radiat Oncol Biol Phys
(2003) - et al.
Alterations in normal liver doses due to organ motion
Int J Radiat Oncol Biol Phys
(2003)
Clinical experience with a commercial respiratory gating system
Int J Radiat Oncol Biol Phys
Reproducibility of organ position with respiratory gating for liver tumorsUse in dose-escalation
Int J Radiat Oncol Biol Phys
Evaluation of the influence of breathing on the movement and modeling of lung tumors
Int J Radiat Oncol Biol Phys
Cited by (0)
Supported by NCI-PO1 Grant 21239 and in part by Deutsche Forschungsgemeinschaft in cooperation with Gesellschaft für Schwerionenforschung (stipend to E.R.).