Respiratory Motion Management for External Beam Radiotherapy

19.1

Introduction

One of the primary goals in radiation therapy is to maximize the dose to the target while minimizing the dose to healthy tissue. For targets that exhibit significant motion (>5 to 10 mm) during the respiratory cycle this creates a challenge. If a target is 3 cm in diameter (volume = 14.1 cm ³ ) and moves ±10 mm in one direction and ±5 mm in the other directions, the volume encompassing the motion becomes an ellipsoid of about 42 cm ³ , which would be the volume needed to ensure that the target is covered throughout the respiratory cycle. This means that approximately 28 cm ³ of normal tissue (200% of the target volume) is irradiated just to account for the target motion.

The motion of organs and tissue also leads to artifacts during the acquisition of treatment planning images. Respiratory-induced blurring can make the delineation of targets and critical structures difficult. In addition to blurring, the interplay of the respiratory cycle and the temporal aspect of the image acquisition can lead to discontinuities in structures or to a complete failure to detect small tumors. If a simulation computed tomography (CT) with such artifacts is used as a reference image during image-guided radiation therapy (IGRT) treatments, major misalignments can occur.

While the preceding discussion certainly applies to the lung, it must be remembered that other treatment sites are affected as well. The liver is strongly affected by respiratory-induced motion and deformation, especially at the dome of the liver. Other organs in the abdomen have been shown to move substantially with respiration (e.g., approximately 7 mm superior-inferior (SI) excursion of the kidneys in an average patient ). This has implications for margins and motion control in sites such as pancreas or Wilms tumor.

Thus the driving forces behind respiratory motion management are to more accurately delineate targets and critical structure and to reduce the volume of normal tissue if possible and thereby reduce complications and/or increase the dose that can be delivered. Therefore methods must be developed to control (or at least account for) respiratory motion. The methods are summarized in Table 19.1 .

Each of these methods has strengths and weaknesses and varying degrees of complexity. Regardless of the method selected, there are some common issues.

• 1.

  The method must undergo rigorous commissioning and ongoing quality assurance (QA).

• 2.

  The respiratory motion must be accounted for during simulation.

• 3.

  The planning margins must be identified.

• 4.

  The respiratory motion during simulation must accurately reflect the respiratory motion during treatment (e.g., if motion control devices are used these must be identical and have the same settings during treatment).

• 5.

  The patient must be able to comply with the motion management strategy.

• 6.

  There may be an additional interplay with respect to intensity-modulated radiation therapy (IMRT) delivery.

• 7.

  Motion management will require more resources and may result in longer treatment times.

• 8.

  Target and normal tissue motion is not limited to the lung. Abdominal structures are also affected.

• 9.

  All imaging modalities are affected by respiratory motion.

• 10.

  Tumor motion is not the only source of uncertainty and may not even be the largest.

The American Association of Physicists in Medicine (AAPM) Task Group (TG)-76, The Management of Respiratory Motion in Radiation Oncology, has produced a comprehensive report detailing each of these strategies. The report is freely available on the AAPM website, and the reader is encouraged to obtain it.

19.2

Measuring Respiratory Motion

The respiratory cycle is initiated by an inhalation caused by the contraction of the diaphragm and the intercostal muscles, which pull the diaphragm down and the ribs anteriorly and superiorly. This increases the size of the thoracic cavity, drawing air into the lungs. At exhale the muscles relax and the thoracic cavity returns to its pre-inhale position. The motion of the diaphragm causes the abdomen to shift inferiorly and anteriorly. The respiration can be characterized as shallow, normal, or deep. During normal inspiration the lung volume can increase by 10% to 25%.

Respiratory motion can be measured by direct observation of the structure of interest, by the use of an implanted fiducial marker, or by a surrogate structure such as the diaphragm or chest wall. Breathing patterns vary from patient to patient, indicating the need for any motion management strategy to be patient specific. Even for the same patient the cycle can vary during a single session or from day to day. Figure 19.1 shows an example of breathing patterns for one patient over a span of minutes. The patient's breathing is regular in the first session (panel A) but becomes more irregular in the second session (panel B). For some patients, audiovisual feedback and respiratory training can be used to help regularize the breathing cycle. This involves displaying the detected respiratory cycle to the patient as the patient breathes, along with visual cues as to the desired amplitudes to “train” the patient to breathe in a regular pattern.

Many imaging modalities have been used to measure organ motion during the respiratory cycle. Table 1 in AAPM TG-76 gives modality-specific references for various organs circa 2006. Table 2 in AAPM TG-76 is a sample of observed motion in the lung from various publications which is adapted and shown in Table 19.2 . Most of the studies quoted involved from 10 to 30 patients. Though the data are representative and were not intended to be exhaustive, several trends are observed. First, the range of motion can be quite large (up to 50 mm in some patients). Second, the maximum range depends on the lobe in the lung, with the lower lobe showing the largest range and the upper lobe the smallest. This intuitively makes sense because the lower lobe is bounded by the diaphragm inferiorly. Third, the SI direction shows the greatest range, followed by the anterior-posterior (AP) direction and then the left-right (LR) direction.

Many studies demonstrated hysteresis in the tumor trajectories. This means that the trajectory during inhalation is different from the trajectory during exhalation and that the maximum range of motion may not be demonstrated at full inhalation. Figure 19.2 shows an example of this. This will have an impact on the motion management strategies discussed in subsequent sections of this chapter.

AAPM TG-76 states that in general abdominal organ motion is in the SI direction with no more than 2 mm range in the AP and LR directions, though this was based on early 4D CT data and is probably not correct (see reference 1). They do qualify that in some individuals the kidneys may display more complex motions.

19.3

Tracking the Respiratory Cycle Through Anatomic Surrogates

With most current technology, it is not possible to directly visualize tumor motion or respiratory motion during the course of treatment (or even during a simulation scan). For this reason it has been important to establish anatomic surrogates to serve as a means of tracking the respiratory cycle. These include the chest wall and abdominal displacement as observed by the change in the patient's external contour. This is achieved by placing a marker on the patient surface and tracking with a camera (infrared or visible light), a pressure-sensitive belt placed around the patient's abdomen, or direct capture and tracking of the patient surface (stereo cameras or lasers). In addition, spirometry and detection of the diaphragm have been used. Table 19.3 shows some of the available technology used to track the respiratory cycle.

With any of these methods it cannot be assumed that there is a direct coupling of the motion of the surrogate anatomy with the tumor itself or that they are in phase with each other. AAPM TG-76 describes many studies in which the phase offset was in the range of 0.5 to 1 second. Furthermore, the phase offset between the surrogate and the tumor itself has been shown to change over the course of treatment. Therefore the use of any surrogate will introduce uncertainty.

19.4

Motion Encompassing

As shown in Table 19.1 , there are several ways to visualize the extent of motion and thereby delineate a volume that encompasses the extent of the motion. Each of these will be discussed in the following sections.

4D CT/Respiratory-Correlated CT

4D CT (or respiration-correlated CT) is a key tool in the measurement and management of respiratory motion. It provides a phase-resolved visualization of tissue motion in 3D, essentially a 3D “movie.” The first paper describing the technique was published in 2003, and vendor adoption was rapid. The technology is now widely available.

To acquire 4D image data sets, knowledge of the point in the respiratory cycle for each projection is required. For this purpose a respiratory trace is acquired (either via a belt or an infrared (IR) marker place on the thorax) during the time when the CT data are acquired. Respiratory sorting is then done retrospectively (correlating each projection with a point in the respiratory cycle) or prospectively (acquiring projections at a defined point in the respiratory cycle). When acquiring a retrospective 4D data set, the respiratory cycle is split into bins, usually 10 bins per breathing cycle, and the slices are sorted by bin. Each bin is then reconstructed into a 3D data set. The 4D motion can then be observed by looping through the 3D images for each bin. In this way the entire range of motion can be visualized. Since there are approximately ten 3D data sets, the labor required to contour would increase by a factor of 10. Vendors have developed algorithms to propagate contours drawn on one set to the rest of the sets to alleviate this situation. Once the automated contours are created they should be reviewed to determine their accuracy. One advantage of 4D CT over slow CT, mentioned above, is that the image in each respiratory bin is quite sharp. There is, however, typically more noise in 4D CT images versus a standard helical scan because vendor protocols typically reduce the mA in 4D CT scans in order to limit the dose.

There is ongoing debate as to how to determine the respiratory bins. This can be done by the phase or by the amplitude. In phase-based binning the bins are determined by their temporal relationship to the cycle, whereas in amplitude-based binning the bins are determined by the fraction of maximum amplitude. Figure 19.3 demonstrates the difference between the two methods. Each method has advantages, although some studies suggest that amplitude-based binning produces fewer or a comparable amount of image artifacts compared to phase-based binning. The algorithm is based on observing if there are irregularities in the amplitude and/or the pattern of the respiratory cycle and then determining the best method based on the combination of the irregularities.

4D images can also be acquired with cone beam CT (CBCT), positron emission tomography (PET), and magnetic resonance (MR) imaging techniques using similar methods. In current technologies, 4D CBCT is acquired by automatically detecting the position of the diaphragm in the planar images acquired during CBCT acquisition.