In veterinary radiation oncology, radiation-planning beam arrangements are usually created by use of static CT images that do not take into account variability in tumor position, of which tumor movement caused by respiration is an important component.1 During the planning process, the treatment planner determines the gross demonstrable extent of the tumor (gross tumor volume) as well as areas thought to contain microscopic disease (clinical target volume) and then adds a 3-D margin to account for variations in tissue location, size, and shape as well as variations in patient position and beam position during and between treatments (intrafractional and interfractional motion, respectively) to calculate planning target volume.1 To determine the size of margins needed to account for intrafractional motion, information about the amount of tumor motion caused by respiration is essential. If the margins used for planning target volume are underestimated and there is movement of the tumor outside of the treatment field, the radiation dose delivered to the clinical target volume will be decreased and risk of treatment failure may be increased. If the required margins are overestimated, the volume of critical organs included in the radiation field can limit the total dose safely deliverable to the clinical target volume.
For human patients undergoing radiation therapy, the American Association of Physicists in Medicine recommends use of respiratory management techniques if movement caused by respiration, measured via a technique such as CT or fluoroscopy, is > 5 mm, with reduction of this value when special procedures such as IMRT and SBRT are performed.2 Respiratory management techniques are not commonly used in veterinary radiation oncology; however, the use of IMRT and SBRT methods is increasing and it may be necessary to incorporate these techniques into curative-intent radiation treatment if the magnitude of respiratory motion exceeds acceptable limits. To our knowledge, there is no information in the veterinary literature on the amount of motion at different anatomic locations caused by respiration. The objectives of the prospective study reported here were to measure the amount of respiratory motion at different locations on the external aspect of the thoracic wall in dogs using a real-time, 3-D motion tracking system and to compare the amount of respiratory motion between dogs positioned with and without a commercially available, vacuum-formable cushion commonly used for veterinary patients. We hypothesized that use of a cushion would decrease the amount of respiratory motion at the external aspect of the thoracic wall.
Materials and Methods
Animals—Eight mixed-breed (Husky cross) research dogs (median age, 3 years [range, 2 to 5 years]; median weight, 23 kg [range, 18.5 to 27.3 kg]) were included in the study. All dogs were determined to be healthy at the start of the study on the basis of results of physical examination. The study was approved by the University of Saskatchewan's Animal Research Ethics Board, and adhered to the Canadian Council on Animal Care guidelines for humane animal use.
Experimental procedures—For respiratory motion measurements, dogs were premedicated with diazepam (0.2 mg/kg, IV) and fentanyl (5 μg/kg, IV), and anesthesia was induced with propofol (4.0 mg/kg, IV) and maintained with sevoflurane (2% to 3% in oxygen, to effect) delivered via a cuffed endotracheal tube. An IV catheter was aseptically placed in a cephalic vein, and an electrolyte-containing solutiona (10 mL/kg/h, IV) was administered throughout anesthesia. Oxygen saturation and Paco2 were continuously monitored, and heart rate, respiratory rate, and systolic, diastolic, and mean arterial blood pressures were recorded every 5 minutes.
Each dog was positioned in dorsal and then in sternal recumbency on a flat, hard surface, first with and then without a vacuum-formable cushion,b during a single anesthetic episode. When dogs were placed in dorsal recumbency, tape was used to secure the forelimbs with the shoulder joints extended and the elbow and carpal joints in a flexed position, and the rear limbs were secured in an extended position. When a cushion was not used for positioning dogs in dorsal recumbency, sandbags were placed on each side of the dog's thorax for stabilization. When dogs were placed in sternal recumbency, the forelimbs and the rear limbs were secured in extended positions. A laser positioning systemc was used to assist with positioning the dogs in true dorsal and sternal recumbency through alignment of various anatomic landmarks, such as the spinous processes of the vertebrae, with the lasers.
Movement of anatomic landmarks on the dogs was measured via a motion capture systemd (nominal accuracy, < 0.3 mm) that tracked the 3-D locations of small (approx 5 mm in diameter) IREDs. The motion capture system was factory calibrated by the manufacturer. Hair was shaved from the skin prior to IRED placement; the laser alignment system was used to guide diode positioning. For each set of measurements, 7 IREDs were placed on each dog. The IREDs placed at the ventral and dorsal thoracic midline were each mounted on the side of a 2 × 2-cm, closed-cell extruded polystyrene foam cube to allow line of sight to the cameras. The IREDs (and supporting cubes, where applicable) were attached directly to the skin surface with cyanoacrylate glue.
When dogs were in dorsal recumbency, 4 equidistant IREDs were placed between the cranial aspect of the manubrium and the caudal aspect of the xiphoid process at midline (markers 1 through 4, with marker 1 placed in the cranialmost position), and 3 IREDs were placed on the right side of the thorax, halfway between the sternum and the vertebrae, at the same cranial-caudal levels as the 3 caudalmost midline markers (markers 5 through 7). Only 3 IREDs were placed on the lateral aspect of the thorax because the line of sight to the camera from a lateral IRED at the cranialmost location was obstructed by the forelimb.
When dogs were in sternal recumbency, 4 IREDs were placed at midline at the levels of T1, T6, T11, and L2 (markers 1 through 4). Positions of the vertebrae were determined via palpation of bony landmarks. Three lateral IREDs were placed at the same cranial-caudal levels as the 3 caudalmost midline IREDs on the left side of the thorax, halfway between the sternum and the vertebrae (markers 5 through 7). When the dogs were in sternal recumbency, the lateral IREDs were placed on the opposite side of the thorax from that used in dorsal recumbency so that a direct line of sight would be present with the detection cameras, which were kept stationary during the measurements.
Three-dimensional movement data were collected for each IRED in the x, y, and z axes (left-to-right, cranial-to-caudal, and ventral-to-dorsal directions, respectively). Movement data were collected for 4 consecutive, 1-minute trials for each experimental condition at a sampling rate of 100 Hz. Data were processed with custom software.e During data collection, each dog was aligned to an orthogonal coordinate system via a fixed laser alignment system.c Orientation of the alignment of the dogs relative to the motion capture system was recorded, allowing the 3-D movement of IREDs to be transformed to correspond with anatomic directions (cranial-caudal, left-right, and ventral-dorsal). Raw coordinate data were low-pass filtered by means of a fourth-order Butterworth filter with a cutoff frequency of 2 Hz and expressed as excursions about the mean value for each axis. A semiautomatic routine was used to identify and extract data for each breath cycle. Peak-to-peak excursions were calculated for each axis of each breath. A measure of overall movement was also generated via calculation of the 3-D Euclidean distance of each point from its mean value over time (ie, displacement vector). The peak Euclidean distance was calculated for each breath (peak displacement vector). The duration of each breath cycle was measured and used to calculate respiratory rate. At the completion of anesthesia, IV catheters were removed and dogs were monitored in a recovery area until able to walk. The IREDs were removed by use of gentle traction prior to recovery.
Statistical analysis—Data were analyzed with a commercially available software program using a linear mixed model.f A random intercept was used to account for dependence in observations associated with multiple types of measurements for each subject (position of the dog, use of a vacuum-formable cushion, and marker locations). An additional repeated term with an autoregressive correlation structure (AR[1]) was also added to the model to account for measurement of multiple breaths for each marker position. The variation in marker displacement associated with differences between dogs was determined in the final model by assessing the proportion of total remaining variance accounted for by the random effect for dog.
Differences in measurements between dorsal and sternal recumbency were determined for the peak displacement vector and then the x, y, and z axes after accounting for respiratory rate and marker position; calculations were performed first for markers 5 through 7 without cushion use and were then repeated with cushion use.
The difference in movement measured as the peak displacement vector with and without a cushion was determined for dogs in dorsal and then sternal recumbency for markers 1 through 4 and then for markers 5 through 7. The model also assessed associations between respiratory rate, body weight, and marker movement. The potential for nonlinear associations between respiratory rate and displacement was examined via assessment of the significance of a polynomial term in the model. Two-way interactions between respiratory rate, marker location, and whether or not a cushion was used were examined and included in the final model when significant.
Finally, the marker positions with the least amount of movement (ie, smallest peak displacement vector) were identified from all observations for dogs in dorsal and then sternal recumbency after accounting for any potential differences associated with cushion use and respiratory rate.
Results
Anesthetic time for all dogs ranged from 65 to 84 minutes, with an approximate time for setup and measurements in each position of 15 to 20 minutes. Least squares mean displacements of the markers for dogs in sternal and dorsal recumbency were similar between those positioned without or with a vacuum-formable cushion (Tables 1 and 2). The differences in displacement for all directions of movement (cranial-caudal, left-right, and ventral-dorsal) and for the peak displacement vector between dogs in sternal and dorsal recumbency were < 0.5 mm for all lateral thoracic markers. Differences were significant (P < 0.05) between dogs in sternal and dorsal recumbency for lateral thoracic markers (markers 5 through 7) in the left-to-right and cranial-to-caudal directions with and without cushion use.
Least squares mean and 95% confidence interval of the mean displacement and mean peak displacement vector of IREDs attached to the thoracic midline (markers 1 through 4) or lateral aspect of the thorax (markers 5 through 7) in 8 healthy mixed-breed dogs positioned without a vacuum-formable cushion during general anesthesia.
| Displacement (mm) | ||||||||
|---|---|---|---|---|---|---|---|---|
| Left-to-right | Cranial-to-caudal | Dorsal-to-ventral | Peak displacement vector (mm) | |||||
| Variable | Mean | 95% CI | Mean | 95% CI | Mean | 95% CI | Mean | 95% CI |
| Sternal recumbency | ||||||||
| Markers 1–4 | 0.65 | 0.32 to 0.98 | 0.64 | 0.35 to 0.93 | 1.37 | 0.89 to 1.85 | 1.40 | 0.86 to 1.94 |
| Markers 5–7 | 1.62 | 0.70 to 2.53 | 0.62 | 0.33 to 0.90 | 2.25 | 1.36 to 3.15 | 2.44 | 1.29 to 3.59 |
| Dorsal recumbency | ||||||||
| Markers 1–4 | 0.84 | 0.51 to 1.18 | 1.21 | 0.92 to 1.49 | 1.50 | 1.03 to 1.98 | 1.67 | 1.13 to 2.21 |
| Markers 5–7 | 1.95 | 1.04 to 2.87 | 0.88 | 0.60 to 1.16 | 2.67 | 1.77 to 3.57 | 2.76 | 1.61 to 3.92 |
| Difference (sternal vs dorsal recumbency) | ||||||||
| Markers 5–7 | −0.33* | −0.67 to −0.01 | −0.27† | −0.47 to −0.06 | −0.41 | −1.00 to 0.18 | −0.33 | −0.83 to 0.18 |
Markers 1 through 4 were located on the dorsal thoracic midline at the levels of T1, T6, T11, and L2 (sternal recumbency) or on the ventral thoracic midline, spaced evenly between the cranial aspect of the manubrium and caudal aspect of the xiphoid process (dorsal recumbency). Markers 5 through 7 were located on the lateral aspect of the thorax, halfway between the sternum and vertebrae, at the same cranial-to-caudal levels as the midline markers. Differences between positions for markers 5 through 7 were derived directly from the model.
Significant differences are indicated P = 0.048
Significant differences are indicated P = 0.01).
CI = Confidence interval.
Least squares mean and 95% confidence interval of the mean displacement and mean peak displacement vector of IREDs attached to the thoracic midline (markers 1 through 4) or lateral aspect of the thorax (markers 5 through 7) in the same 8 dogs in Table 1 positioned with a vacuum-formable cushion.
| Displacement (mm) | ||||||||
|---|---|---|---|---|---|---|---|---|
| Left-to-right | Cranial-to-caudal | Dorsal-to-ventral | Peak displacement vector (mm) | |||||
| Variable | Mean | 95% CI | Mean | 95% CI | Mean | 95% CI | Mean | 95% CI |
| Sternal recumbency | ||||||||
| Markers 1–4 | 0.39 | 0.23 to 0.55 | 0.78 | 0.38 to 1.19 | 1.77 | 1.33 to 2.21 | 1.58 | 1.10 to 2.06 |
| Markers 5–7 | 1.02 | 0.50 to 1.53 | 0.51 | 0.33 to 0.69 | 2.18 | 1.47 to 2.89 | 2.06 | 1.27 to 2.85 |
| Dorsal recumbency | ||||||||
| Markers 1–4 | 0.59 | 0.43 to 0.75 | 1.29 | 0.89 to 1.69 | 1.37 | 0.93 to 1.81 | 1.46 | 1.10 to 2.06 |
| Markers 5–7 | 1.51 | 0.99 to 2.03 | 0.77 | 0.58 to 0.94 | 2.21 | 1.50 to 2.93 | 2.24 | 1.45 to 3.04 |
| Difference (sternal vs dorsal recumbency) | ||||||||
| Markers 5–7 | −0.50* | −0.77 to −0.22 | −0.26* | −0.39 to −0.13 | −0.03 | −0.49 to 0.42–0.18 | −0.60 to 0.23 | |
Differences are significant (P < 0.001).
See Table 1 for remainder of key.
Differences among dogs accounted for a substantial proportion of the overall variation for mean displacements in all directions and for peak displacement vector. The amount of total variance of the peak displacement vector associated with differences among dogs, after accounting for respiratory rate, marker, position (dorsal vs ventral), vacuum-formable cushion use, and repeated measurements, was 45.6% for markers 1 through 4 and 67.8% for markers 5 through 7 when dogs were positioned in sternal recumbency and was 32.8% for markers 1 through 4 and 40.1% for markers 5 through 7 when dogs were positioned in sternal recumbency.
For dogs in dorsal recumbency, differences in the peak displacement vector for ventral thoracic midline markers 1 through 4 were nonsignificant (P = 0.11) between those positioned with or without a vacuum-formable cushion, and findings were similar for lateral thoracic markers 5 through 7 (P = 0.19; Tables 1 and 2). For dogs in sternal recumbency, there was also no significant (P = 0.41) difference in the peak displacement vector for lateral thoracic markers 5 through 7 between dogs positioned with and without a cushion. There were, however, significant (P < 0.001) differences detected in peak displacement vector for 3 of the 4 dorsal thoracic midline markers between dogs positioned in sternal recumbency with and without a cushion (Table 3).
Least squares mean and 95% confidence interval of peak displacement vectors for IREDs attached to the dorsal thoracic midline (markers 1 through 4) of the same 8 dogs in Table 1 positioned in sternal recumbency with and without a vacuum-formable cushion and differences between values for each marker.
| Peak displacement vector (mm) | |||||||
|---|---|---|---|---|---|---|---|
| With cushion | Without cushion | Difference | |||||
| Marker | Mean | 95% CI | Mean | 95% CI | Mean | 95% CI | P value |
| 1 | 1.02 | 0.43–1.60 | 0.88 | 0.30–1.47 | 0.14 | −0.05 to 0.32 | 0.57 |
| 2 | 1.53 | 0.94–2.11 | 1.26 | 0.67–1.84 | 0.27 | 0.08 to 0.45 | < 0.001 |
| 3 | 1.99 | 1.40–2.57 | 1.60 | 1.02–2.19 | 0.38 | 0.20 to 0.57 | < 0.001 |
| 4 | 2.34 | 1.75–2.92 | 1.83 | 1.24–2.41 | 0.51 | 0.33 to 0.70 | < 0.001 |
See Table 1 for key.
The median respiratory rate for all dogs was 11.7 breaths/min (range, 3.6 to 21.1 breaths/min). For dogs in sternal recumbency, there was no significant difference in peak displacement vector associated with respiratory rate for markers 1 through 4 (P = 0.10) or for markers 5 through 7 (P = 0.38). For all dogs in dorsal recumbency, an increase in respiratory rate was associated with a mean ± SE decrease of 0.022 ± 0.003 mm (P < 0.001) in peak displacement vector for markers 1 through 4 and 0.029 ± 0.002 mm (P < 0.001) for markers 5 through 7. The association between respiratory rate and peak displacement vector for markers 1 through 4 was not significantly (P = 0.14) affected by use of a vacuum-formable cushion for dogs in dorsal recumbency. However, the association between respiratory rate and peak displacement vector for markers 5 through 7 was significantly (P = 0.006) affected by cushion use in the same position; an increase in respiratory rate was associated with a mean ± SE decrease of 0.030 ± 0.002 mm (P < 0.001) in the peak displacement vector when a cushion was used, whereas no association was detected between respiratory rate and peak displacement vector when no cushion was used (P = 0.62). There were no significant nonlinear trends in respiratory rate-associated marker displacement among dogs (P = 0.093 and 0.66 for markers 1 through 4 and markers 5 through 7, respectively).
Comparisons among individual dorsal midline markers for all dogs in sternal recumbency revealed that the peak displacement vector was significantly (P < 0.001) lower for marker 1 than for markers 2 through 4. In dorsal recumbency, the peak displacement vectors for markers 1 and 2 were significantly (P < 0.01 for both comparisons) lower than for markers 3 and 4. For dogs in sternal or dorsal recumbency, the peak displacement vector was lower for marker 6 than for markers 5 and 7 (P = 0.02). There was no significant (P ≥ 0.66 for all comparisons) association between the peak displacement vector and weight for dogs in any position, with or without a cushion, for any markers.
Discussion
In the present study, 3-D movement of the dorsal, ventral, and lateral aspects of the thoracic wall in 8 healthy mixed-breed dogs was measured by use of 7 externally attached IREDs monitored via a motion capture system. Four markers were located on the dorsal thoracic midline at the levels of T1, T6, T11, and L2 (for dogs in sternal recumbency) or on the ventral thoracic midline, spaced evenly between the cranial aspect of the manubrium and caudal aspect of the xiphoid process (for dogs in dorsal recumbency). Three markers were located on the lateral aspect of the thoracic wall, halfway between the sternum and vertebrae, at the same cranial-to-caudal levels as the midline markers; these were placed on the left and right sides for dogs in sternal and dorsal recumbency, respectively.
Placement of anesthetized dogs in sternal recumbency significantly decreased respiratory motion at the lateral aspect of the thoracic wall in the left-to-right and cranial-to-caudal directions, whether or not a vacuum-formable positioning cushion was used. This decrease in movement most likely resulted from compression of the ventral thoracic region by gravity, potentially restricting lateral expansion of the thorax. Although there were significant differences in the amount of movement at the lateral aspect of the thoracic wall in some directions between dogs in sternal and dorsal recumbency, the magnitude of the differences was considered small (≤ 0.5 mm), and the peak displacement vector (measurement of overall motion in 3-D space) was not different between positions. For this reason, other factors such as repeatability of positioning and ease of patient positioning for radiation treatment may be of greater importance when deciding between dorsal and sternal positioning for treatment of a patient. The weight, positioning, and conformation of dogs used in this study as well as the anesthetic protocols used should be considered when evaluating the results and considering their applicability to other dogs undergoing radiation therapy. It is possible that these variables may affect the amount of respiratory movement. Although no association between weight and the peak displacement vector was found, our study included a small number of healthy dogs with a narrow range of weights (18.5 to 27.3 kg). Furthermore, the findings are applicable to the superficial aspects of the thoracic wall only, as there may not be a direct correlation between movement of the external surface and deeper structures. For example, movements of skin fiducial markers may not correlate with lung motion.3
For dogs in sternal recumbency, use of a vacuum-formable positioning cushion significantly increased respiratory motion at 3 of the 4 dorsal thoracic midline markers, suggesting that a cushion should only be used if the benefits, such as improved repeatability of positioning, outweigh this effect. The increase in motion was most likely caused by the cushion restricting thoracic wall expansion in a lateral direction and shifting respiratory motion to the dorsal-to-ventral direction.
Previous studies of anesthetized dogs in dorsal recumbency have reported the greatest motion of the chest wall in the lateral dimension,4,5 which contrasts with our findings of maximum mean displacement in the dorsal-to-ventral direction (no statistical test performed). The use of a positioning cushion in the present study, which restricted lateral motion and increased movement of the dorsal thoracic midline for dogs in sternal recumbency, may have led to these differences, as could differences in size and breed of dogs used in the present and previous studies. Additionally, dogs in the present study were assessed without cushions in sternal and dorsal recumbency, in contrast to the previous studies that included dogs in dorsal recumbency only.
Increased respiratory rate resulted in a significant decrease in motion of all markers for dogs in dorsal recumbency. However, this association between breathing rate and the peak displacement vector was lost for lateral thoracic wall markers when no cushion was used. The effect of respiratory rate on magnitude of marker movement would be expected; a higher respiratory rate may be associated with a shallower depth of respiration, resulting in less respiratory motion. A technique such as rapid, shallow, manual ventilation could be applied to veterinary patients quite easily and with low cost. In human patients, this technique involves the application of abdominal pressure to reduce diaphragmatic excursions. Veterinary patients, unlike adult human patients, commonly receive radiation treatments under general anesthesia, and a shallow depth of respiration could also be achieved with ventilator-controlled breathing.
The marker locations with the least amount of respiratory motion in the present study could potentially be considered for placement of skin markings used in preparation of canine patients for radiation therapy to decrease positioning uncertainty. These locations included the level of T1 for dorsal midline markers, the cranial third of the region between the manubrium and the xiphoid process for ventral midline markers, and the approximate level of the T6 for markers on the lateral aspect of the thorax. Respiratory motion varied markedly among dogs in our study, as has been reported in human patients.2 Individual assessment of respiratory motion prior to treatment is recommended where possible for human patients; however, use of a set margin or respiratory management methods are more practical options for veterinary patients at this time. The real-time motion capture system used in the present study allowed easy and minimally invasive measurement of respiratory motion in 3 dimensions. However, line of sight is needed between the IREDs and the camera, which limits its use to external anatomy. As well as application to anatomic locations other than the thoracic wall, this system could potentially be used to explore the relationship between external surface motion and internal tumor motion in future studies.
The findings of the study reported here provide a basis for further research on respiratory motion in anesthetized dogs, and consideration of these results may be useful when determining margin selection in dogs with characteristics similar to those of dogs used in the present study, treated under similar conditions. Motion correction for conventional radiation therapy may not be necessary if additional margins, determined on the basis of the magnitude of respiratory motion, can be added to compensate for potential movement. However, for highly conformal treatments such as IMRT and SBRT, expansion of treatment margins to account for respiratory tumor motion will increase the volume of normal tissues exposed to a high dose, and results of the present study indicate that respiratory management techniques such as shallow mechanical ventilation should be considered, particularly if a low number of fractions are used.
ABBREVIATIONS
| IMRT | Intensity-modulated radiation therapy |
| IRED | Infrared light–emitting diode |
| SBRT | Stereotactic body radiotherapy |
Normosol-R, Hospira, Montreal, QC, Canada.
Vac-Lok, Civco Medical Solutions, Orange City, Iowa.
LAP Dorado CT4–1-POST, LAP of America LC, Boynton Beach, Fla.
VZ-3000, Visualeyez, Phoenix Technologies Inc, Burnaby, BC, Canada.
Matlab R2006b, Mathworks Inc, Natick, Mass.
PROC MIXED, SAS, version 9.2 for Windows, SAS Institute Inc, Cary, NC.
References
- 1.↑
International Commission on Radiation Units and Measurements. Report 62: prescribing, recording and reporting photon beam therapy (supplement to ICRU Report 50). Bethesda, Md: International Commision on Radiation Units and Measurements Inc, 1999.
- 2.↑
Keall PJMageras GSBalter JM, et al. The management of respiratory motion in radiation oncology report of AAPM Task Group 76. Med Phys 2006; 33: 3874–3900.
- 3.↑
Koch NLiu HHStarkschall G, et al. Evaluation of internal lung motion for respiratory-gated radiotherapy using MRI: part I—correlating internal lung motion with skin fiducial motion. Int J Radiat Oncol Biol Phys 2004; 60: 1459–1472.
- 4.
Warner DOKrayer SRehder K, et al. Chest wall motion during spontaneous breathing and mechanical ventilation in dogs. J Appl Physiol 1989; 66: 1179–1189.
- 5.
Schmid ERRehder KKnopp TJ, et al. Chest wall motion and distribution of inspired gas in anesthetized supine dogs. J Appl Physiol 1980; 49: 279–286.
