- Email: [email protected]
Ultrasound-guided extracranial radiosurgery
Int. J. Radiation Oncology Biol. Phys., Vol. 55, No. 4, pp. 1092–1101, 2003 Copyright © 2003 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/03/$–see front matter
doi:10.1016/S0360-3016(02)04406-1
PHYSICS CONTRIBUTION
ULTRASOUND-GUIDED EXTRACRANIAL RADIOSURGERY: TECHNIQUE AND APPLICATION SANFORD L. MEEKS, PH.D.,* JOHN M. BUATTI, M.D.,* LIONEL G. BOUCHET, PH.D.,† FRANCIS J. BOVA, PH.D.,† TIMOTHY C. RYKEN, M.D.,‡ EDWARD C. PENNINGTON, M.S.,* KATHLEEN M. ANDERSON, C.M.D.,* AND WILLIAM A. FRIEDMAN, M.D.† Departments of *Radiation Oncology and ‡Neurosurgery, University of Iowa, Iowa City, IA; †Department of Neurological Surgery, University of Florida, Gainesville, FL Purpose: Stereotactic radiosurgery is an effective treatment modality for many intracranial lesions, but target mobility limits its utility for extracranial applications. We have developed a new technique for extracranial radiosurgery based on optically guided three-dimensional ultrasound (3DUS). The 3DUS system provides the ability to image the target volume and critical structures in real time and determine any misregistration of the target volume with the linear accelerator. In this paper, we describe the system and its initial clinical application in the treatment of localized metastatic disease. Methods and Materials: The extracranial stereotactic system consists of an ultrasound unit that is optically tracked and registered with the linear accelerator coordinate system. After an initial patient positioning based on computed tomographic (CT) simulation, stereotactic ultrasound images are acquired and correlated with the CT-based treatment plan to determine any soft-tissue shifts between the time of the planning CT and the actual treatment. Optical tracking is used to correct any patient offsets that are revealed by the real-time imaging. Results: Preclinical testing revealed that the ultrasound-based stereotactic navigation system is accurate to within 1.5 mm in comparison with an absolute coordinate phantom. Between March 2001 and March 2002, the system was used to deliver extracranial radiosurgery to 17 metastatic lesions in 16 patients. Treatments were delivered in 1 or 2 fractions, with an average fractional dose of 16 Gy (range 12.5–24 Gy) delivered to the 80% isodose surface. Before each fraction, the target misalignment from isocenter was determined using the 3DUS system and the misalignments averaged over all patients were anteroposterior ⴝ 4.8 mm, lateral ⴝ 3.6 mm, axial ⴝ 2.1 mm, and average total 3D displacement ⴝ 7.4 mm (range ⴝ 0 –21.0 mm). After correcting patient misalignment, each plan was delivered as planned using 6 –11 noncoplanar fields. No acute complications were reported. Conclusions: A system for high-precision radiosurgical treatment of metastatic tumors has been developed, tested, and applied clinically. Optical tracking of the ultrasound probe provides real-time tracking of the patient anatomy and allows computation of the target displacement before treatment delivery. The patient treatments reported here suggest the feasibility and safety of the technique. © 2003 Elsevier Science Inc. Extracranial stereotactic radiosurgery, 3D ultrasound, Image guidance, Optical tracking, Metastases, Patient positioning.
INTRODUCTION
tached to the ring. The image coordinates are transformed into stereotactic coordinates through knowledge of the fiducial geometry, and hence the stereotactic coordinates of the target volume are determined. Radiation treatment plans are generated based on the three-dimensional (3D) digital image data, and are followed by positioning the actual patient in the treatment vault by using the stereotactic coordinates mathematically determined relative to the stereotactic ring. This rigid localization system allows extremely accurate targeting of the radiation beam; a dose distribution can be
Stereotactic radiosurgery was defined by Lars Leksell as single-fraction irradiation of intracranial targets that could replace open surgery for selected patients. Historically, this procedure has been performed using an invasive stereotactic head frame. The frame is attached to the patient under local anesthesia using four aluminum pins. This ring system provides a rigid frame of reference for all imaging and target localization. Subsequent to head ring placement, images are acquired using a system of fiducials that are at-
Acknowledgment—Development of the ultrasound navigational system was performed in collaboration with ZMed, Inc., Ashland, MA. Received Jul 19, 2002, and in revised form Oct 31, 2002. Accepted for publication Nov 11, 2002.
Reprint requests to: Sanford L. Meeks, University of Iowa, Department of Radiation Oncology, W189Z-GH, 200 Hawkins Dr., Iowa City, IA 52242-1077. Tel: (319) 356-0881; Fax: (319) 356-1530; E-mail: [email protected] Funding for this work was provided by Whitaker Foundation Biomedical Engineering Research Grant RG-98-0278. 1092
Extracranial radiosurgery
generated that conforms very tightly to the target volume while delivering a very small fraction of that dose to the normal tissues just a few millimeters away. To accomplish this steep dose gradient, multiple, noncoplanar beams of radiation are stereotactically directed on the target. Advantages of this minimally invasive form of treatment for brain tumors are that it is a single outpatient treatment, and that radiation exposure to normal brain tissue is reduced and hence many of the complications associated with conventional radiation therapy are avoided. Although the rigid stereotactic fixation enables accurate radiation treatment, the head ring has limited the procedure to the treatment of intracranial lesions. Because of the success of stereotactic radiosurgery for many intracranial applications, there has been an interest in extending the technology to other treatment sites. For example, several groups have used standard head ring localization to treat primary and recurrent head-and-neck cancers that are confined to the base of skull and upper cervical spine (1–9). Several methods have been proposed to further extend stereotactic localization to extracranial radiosurgery (10 –16). High-precision extracranial radiation delivery is challenging, however, because the target position can shift relative to bony anatomy between the time of image acquisition and the time of treatment. Therefore, several groups are actively investigating real-time imaging at the time of treatment as a mechanism for improved high-precision radiotherapy localization (17–20). Our group has recently investigated the use of optically guided 3D ultrasound for high-precision radiation delivery (21–23). The purpose of this paper is to report on the initial clinical use of the system for radiosurgery of metastatic extracranial lesions. METHODS AND MATERIALS Optical guidance Three-dimensional treatment planning in radiation therapy requires acquisition of 3D image sets, such as computed tomography (CT), magnetic resonance imaging, or positron emission tomography, relative to some fiducial system. A virtual simulation of the patient’s treatment is then performed using these 3D image data. In other words, a model of the patient is created in a computerized treatment planning system using the imaging data. The placement of radiation beams is modeled on the virtual patient, and radiation dose distributions are calculated and displayed on the images. Radiosurgery is unique in that it relies on 3D, or stereotactic, image localization with a robust fiducial system, thereby enabling co-identification of the virtual target in the treatment-planning computer with the actual target position for treatment delivery. To help minimize errors, traditional radiosurgery uses a minimally invasive head ring to coordinate the virtual and real worlds and to minimize motion during image acquisition and treatment delivery. We have previously described a system for noninvasive localization of intracranial targets based on optical tracking of a dental tray attached to the patient’s maxillary dentition
● S. L. MEEKS et al.
1093
(24 –26). Optical tracking is simply a means of determining in real time the position of an object by tracking the positions of either active or passive infrared markers attached to the object. The position of the point of reflection is determined using a camera system. In the commercial version of this system (RadioCameras, ZMed, Inc., Ashland, MA), patient localization is accomplished through detection of four passive markers attached to a custom bite plate that links to the maxillary dentition of the patient to form a rigid system (Fig. 1a). This tracking is accomplished in real time using a charge-coupled device (CCD) optical system that is equipped with infrared illuminators that flood the room with infrared light. The infrared light is reflected off of the four passive markers, and these reflections are read using the CCD optics. The position of the markers is then input to a computer system that combines this information with the stereotactic CT localization of the patient (based on this same dental tray fiducial system) and a calibration matrix that relates the camera position to the linac isocenter. The output from the computer is the patient’s displacement from the desired position (i.e., target at linac isocenter), thereby tracking the patient relative to isocenter (Fig. 1b). The ability of the tracking to occur continuously with updated localization approximately 15 times per second makes realtime positioning and patient monitoring practical. Although this system was originally designed for fractionated stereotactic radiotherapy, it has proven to provide extremely accurate localization relative to the treatment room, and has therefore been used for single-fraction radiosurgery in a select cohort of patients (27). Optically guided ultrasound for extracranial radiosurgery System description. While the initial system for opticalguided radiotherapy provides a submillimeter localization accuracy (25–27), it has been limited to intracranial therapy. Outside of the cranium, soft-tissue targets can move relative to rigid fixation points (e.g., bone structures) between the times of image acquisition, treatment planning, and treatment delivery. Therefore, real-time imaging is required to establish extracranial stereotactic localization of the lesion at the time of treatment delivery. We have developed a system for 3D ultrasound guidance (commercial version SonArray, ZMed, Inc., Ashland, MA) to correct for these misalignments at the time of treatment. Ultrasound was chosen because it is an inexpensive, yet flexible and highresolution imaging modality that can easily be adapted for use in a radiation therapy treatment room. The interpretation of two-dimensional ultrasound images is difficult, however, and can be highly dependent on the skill and expertise of the operator in manipulating the transducer and mentally transforming the 2D images into a 3D tissue structure. Much of this difficulty results from using a spatially flexible 2D imaging technique to view 3D anatomy. Three-dimensional ultrasound reconstruction helps overcome this limitation. We generate 3D ultrasound data sets through optical tracking of free-hand acquired 2D ultrasound data. The operator holds the ultrasound probe and manipulates it over the
1094
I. J. Radiation Oncology
● Biology ● Physics
Volume 55, Number 4, 2003
Fig. 1. (a) The intracranial optical-guided radiotherapy system relies on an array of four passive markers attached to the maxillary dentition using a custom bite plate. Infrared reflections off of the passive markers are read by CCD optics, then input to a computer system which then determines the patient’s position based on these reflections. (b) The computer monitor outputs information about the patient’s displacement from isocenter. The top three numbers indicate the patient’s translational misalignments in millimeters along three principal axes, whereas the bottom three numbers indicate the rotational misalignments in degrees about each of these three axes. The middle number, labeled “vector,” is the root mean square of the three translational misalignments, and is therefore equal to the 3D error. The numbers update in real time, allowing the treatment technologist to correct the misalignments.
anatomic region of interest. The raw 2D data are transferred to a computer workstation using a standard video link. The position and angulation of the ultrasound probe in any arbitrary orientation is determined using an array of four
infrared light-emitting diodes (IRLEDs) attached to the probe (Fig. 2a). Similar to our system for intracranial optic guidance, CCD cameras are used to determine the positions of the IRLEDs, and this information is input to the computer
Extracranial radiosurgery
● S. L. MEEKS et al.
1095
Fig. 2. (a) The ultrasound probe is tracked via an array of four IRLEDs rigidly attached to it. A rocking cradle is used to help guide the free-hand acquisition and to ensure that no gaps exist in the final 3D volume. (b) By tracking the probe, an ultrasound volume can be generated. Shown are three orthogonal views of a transabdominal prostate scan.
workstation. The position of each ultrasound pixel can therefore be determined using the IRLEDs, and an ultrasound volume can be reconstructed by coupling the position information with the raw ultrasound data (Fig. 2b). In addition to building the 3D image volume, image
optical guidance is used to determine the absolute position of the ultrasound volume in the treatment room coordinate system. Because the relative positions of the ultrasound volume and the ultrasound probe are fixed, the knowledge of the probe position in the treatment room coordinate
1096
I. J. Radiation Oncology
● Biology ● Physics
system at the time of data acquisition is sufficient to determine the position of the image volume relative to the linac isocenter. The determination of the relative position of the image and probe corresponds to a calibration step that is performed at the time of system installation (21). Preclinical testing. The accuracy of optically guided 3D ultrasound for extracranial radiosurgery patient positioning was tested using a specially designed absolute ultrasound phantom (21, 23). This phantom consists of 15 echoic spheres imbedded in a tissue-equivalent nonechoic medium. The spheres are arranged in sets of five located at three different depths: 30, 60, and 130 mm. A CT scan (0.51 mm ⫻ 0.51 mm ⫻ 1.25 mm) of the phantom was acquired with a passive infrared tracking array attached to it. An infrared array was used to track the position of the phantom in the room coordinate system. Then, the ultrasound phantom was placed in the treatment vault and the ultrasound probe was fixed on top of the phantom. Coupling of the probe and the phantom was achieved with a thin layer of water. Next, a sphere within the phantom was selected as the target sphere and positioned at the room isocenter based on the CT-optical tracking of the phantom. Next, the localization of the target sphere was determined using our 3D ultrasound guidance system. The 3D ultrasound volume of the sphere was acquired using optical tracking as described above. The 3D ultrasound– based position of the target sphere was determined by finding the center of the sphere in the image using a circle tool placed on each three-orthogonal ultrasound views. The target localization accuracy using the 3D ultrasound optic guided system was thus determined by comparing its experimentally determined position to the position of the target sphere as determined from the CT scan. More complete details regarding preclinical testing, commissioning, and quality assurance of the optically guided ultrasound system can be found in the literature (21, 23, 28). Clinical application. Clinical use of ultrasound image guidance proceeds as follows. Before CT scanning, the patient is immobilized using a custom vacuum cushion as is commonly used in radiation therapy (Vac-Loc, Med-Tec, Inc., Orange City, IA). The CT is acquired with the patient immobilized in the same position that will be used during the radiotherapy treatment to maintain a generally consistent position of mobile anatomy. The CT images are transferred to our 3D treatment planning system (Pinnacle3, ADAC, Milipitas, CA), where the tumor volume and normal structures of interest are delineated. A treatment plan is then designed to conform the prescription dose closely to the planning target volume (PTV), while minimizing the dose to the nearby normal structures. This is achieved by using maximally separated noncoplanar beams. Maximal beam separation provides unique beam entry and exit pathways for each beam, thereby optimizing the dose gradient outside of the target volume. In addition, maximal beam separation provides the most unique beam’s-eye views (BEV) of the target volume, thereby maximizing the plan conformality. This is an extension of geometric planning
Volume 55, Number 4, 2003
Fig. 3. Diagrammatic representation of the beam geometries available for treatment planning.
techniques used in intracranial stereotactic therapy (29), with the primary difference being that intracranial radiotherapy planning allows nearly unrestricted beam entrance over 2 space (i.e., the top half of the patient’s head). Due to possible collisions of the gantry with the patient or the treatment couch, the available geometry for beam entrance is much more limited in extracranial planning. The generally available geometry is illustrated in Fig. 3. When the treatment couch is at its home position (0°, International Electrotechnical Commission angle convention), the gantry is capable of full rotation about the patient. As the couch is moved away from its home position, the gantry is capable of rotating approximately ⫾30° from vertical, creating a cone of possible beam entrances above the patient. As a starting point, the gantry and treatment table angles are chosen to achieve maximal beam separation within this limited space of achievable treatment geometries. The planner then inspects the beam’s eye views for each of these beams and modifies the gantry and table angles to avoid critical structures or minimize the projection of the PTV. Each beam is shaped to match the BEV projection of the planning target volume using a multileaf collimator (MLC) that has a 5-mm leaf resolution at isocenter (Millennium MLC-120, Varian Oncology Systems, Palo Alto, CA). The field shapes are designed with zero margin added to the BEV projection of the PTV, and MLC leaves at the edge of the target. The day of the treatment, the patient is placed in the same immobilization cushion that was used during CT scanning. The patient is initially set up relative to isocenter using conventional laser alignment. A 3D ultrasound volume is then acquired and reconstructed in the computer workstation. The target volume and critical structure outlines, as delineated on the planning CT scans, are overlaid on the acquired ultrasound volume in relation to isocenter. The
Extracranial radiosurgery
contours determined from the CT scans are then manipulated until they align with the anatomic structures on the ultrasound images. The amount of movement required to align the contours with the ultrasound images determines the magnitude of the target misregistration with isocenter based on conventional setup techniques. The target is then placed at the isocenter by tracking an infrared array attached to the treatment couch, which allows precise translation from the initial position to the 3D ultrasound determined position. Once all of the setup information has been verified using repeat ultrasound acquisition and coregistration, treatment proceeds as planned. Initial clinical experience Between March 2001 and March 2002, 16 patients with 17 extracranial metastatic lesions have been treated with ultrasound-guided stereotactic radiosurgery at the University of Iowa. Institutional Review Board approval was obtained for review of these patient treatments. Patient characteristics are shown in Table 1. The patients ranged in age from 51 to 79, with a mean age of 64. There were 10 males and 6 females. Treatment sites were liver/porta hepatis (4), paravertebral/t-spine (3), sacral (2), low neck (2), chest wall (2), hip (1), iliopsoas muscle (1), and arm (1). Eleven of the 16 patients had prior conventional irradiation to the treatment area either for treatment of primary disease or initial presentation of metastatic disease. RESULTS Absolute accuracy of 3D ultrasound image guidance As described earlier, the room positions of the 15 spheres in the absolute phantom were determined experimentally using the 3D ultrasound system, and compared against their known positions. Averaging the difference between the measured and expected for all 15 spheres at all depths, the
● S. L. MEEKS et al.
1097
Table 2. Accuracy of image-guided 3D ultrasound as a function of the depth of the target Depth (mm)
Anteroposterior distance (mm)
Lateral distance (mm)
Axial distance (mm)
30 60 130 All depths
0.4 ⫾ 0.2 0.1 ⫾ 0.1 0.6 ⫾ 0.1 0.3 ⫾ 0.4
0.6 ⫾ 0.1 0.7 ⫾ 0.2 1.4 ⫾ 0.4 0.9 ⫾ 0.4
0.2 ⫾ 0.3 0.5 ⫾ 0.3 0.3 ⫾ 0.4 0.3 ⫾ 0.2
full accuracy of this system for patient positioning was found to be on average anteroposterior (AP) ⫽ 0.3 mm, lateral ⫽ 0.9 mm, axial ⫽ 0.3 mm. More detailed analysis can be found in Table 2, but over all depths the accuracy of target repositioning using our ultrasound image-guided system was maintained below 1.5 mm, and generally below 1.0 mm. Initial clinical experience Treatment planning data for the 16 initial patients can be found in Table 3. The average lesion volume was 75.6 cc (range 3.9 –197.3 cc). A conformal treatment plan was generated for each patient; the average number of beams used was 9.1, with a range from 6 to 11 noncoplanar beams. A typical extracranial radiosurgery treatment plan is shown in Fig. 4, illustrating the conformality and dose gradients achievable using this planning methodology. The prescription isodose volume (i.e., the dose prescribed to the lesion periphery) was chosen such that ⱖ95% of the PTV was inside of the prescription isodose volume. The average prescription isodose to target volume ratio (PITV) was 1.75 (range 1.17–2.69). Fourteen of the 16 patients received single-fraction treatment; for these patients, the mean treatment dose to the lesion periphery was 16.4 Gy, with a range of 12.5–24 Gy. The two remaining patients each received
Table 1. Characteristics of patients treated with extracranial radiosurgery Patient No.
Metastatic lesion location
Primary tumor
Age
Previous dose to treatment region (Gy)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Low neck Lung (chest wall) Hip Liver (2 lesions) Paravertebral (T12-L1) T spine Liver Iliopsoas muscle Porta hepatis Neck Sacral Liver Sacral Arm Liver/Rib T spine
Neck Lung Lung Breast Lung Esophagus Colorectal Spindle cell Ovarian Anaplastic thyroid Renal cell Breast Hepatocellular SCCA skin Hepatocellular Esophagus
79 68 51 57 55 65 76 79 51 67 70 58 65 58 53 66
50 60 30 No 66.6 50.4 No No No 30 42.5 No 30 30 30 50.4
SCCA ⫽ squamous cell carcinoma.
1098
I. J. Radiation Oncology
● Biology ● Physics
Volume 55, Number 4, 2003
Table 3. Characteristics of extracranial radiosurgery treatment plans Patient No.
Volume (cc)
Fields
Dose/fraction (Gy)
Fractions
Isodose line (%)
PITV
1 2 3 4
77.2 51.1 19.6 82.4 19.9 81.2 26.9 125.3 9.6 156.6 81.5 197.3 3.9 20.9 83.2 161.3 88.0
7 11 11 11 11 9 7 9 11 9 6 7 11 9 7 9 10
15.0 17.5 17.5 15.0 15.0 15.0 12.5 12.5 24.0 12.5 12.5 17.5 20.0 17.5 17.5 17.5 12.5
1 1 1 1 1 1 1 2 1 2 1 1 1 1 1 1 1
80 80 80 80 80 80 80 90 80 80 80 80 80 80 80 80 70
1.17 1.36 1.60 1.43 1.54 1.53 1.63 1.65 2.00 1.79 1.71 1.63 2.69 2.06 1.62 1.72 2.60
5 6 7 8 9 10 11 12 13 14 15 16
PITV ⫽ prescription isodose to target volume ratio.
25.0 Gy in 2 fractions of 12.5 Gy each. The average prescription isodose surface was 80% (range 70 –90%), normalized to the point of maximum dose. Although this choice of dose prescription results in a significant dose inhomogeneity when compared with conventional radiotherapy, the 80% prescription isodose was chosen for consistency with intracranial radiosurgery (30). The images and treatment plan information were then transferred to the SonArray treatment guidance station. The patient was placed on the treatment couch, and was initially positioned using conventional laser positioning. An optically guided 3D ultrasound volume was acquired, and the contours of the tumor volume and critical structures, as determined in the planning CT scan, were overlaid on the ultrasound volume. The outlines were manipulated until the best fit between the contours and the ultrasound volume was obtained, as illustrated in Fig. 5. The target misalignments from isocenter averaged over all patients were AP ⫽ 4.8 mm, lateral ⫽ 3.6 mm, axial ⫽ 2.1 mm, and average total 3D displacement ⫽ 7.4 mm (range ⫽ 0 –21.0 mm). Data for
all patients are shown in Table 4. Using optical guidance, the patient misalignment was corrected. A repeat ultrasound acquisition and registration was performed to confirm any patient shift, and treatment was accomplished as planned. No acute complications were reported. DISCUSSION Radiosurgery has dramatically changed the treatment for many central nervous tumors including metastatic lesions, but the need for high-precision treatment delivery has limited its utility primarily to intracranial applications. Extracranial application of high-precision radiosurgery is more complex because motion of soft-tissue targets relative to bony references between the time of image acquisition and treatment delivery is unavoidable. We have developed a system that relies on 3D ultrasound optical guidance to precisely correct for internal organ motion, and adjust the position of the target at the isocenter of the treatment. Because our image-guided system uses 3D images, it has
Fig. 4. The treatment isodose distribution is shown superimposed on axial, sagittal, and coronal CT images through the center of the volume. The center isodose line surrounding the PTV is the treatment line (80%), with the 40% and 16% lines also shown.
Extracranial radiosurgery
● S. L. MEEKS et al.
Fig. 5. The basic screen layout for SonArray shows three orthogonal views through the ultrasound volume with the CT contours superimposed. (a) SonArray layout for targeting an iliopsoas metastasis. (b) SonArray layout for targeting a sacral metastasis.
1099
1100
I. J. Radiation Oncology
● Biology ● Physics
Table 4. Misalignment of target volume from isocenter, as measured using optically guided 3D ulstrasound Shift (mm) Patient No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
AP
Lateral
Axial
3D (root mean square)
4.0 ⫺8.2 0.8 ⫺3.6 3.0 ⫺3.9 3.7 3.0 0.0 0.0 ⫺4.0 19.0 0.0 ⫺13 0.0 11.0
3.0 ⫺5.6 0.5 4.7 ⫺11.6 0.0 ⫺3.0 ⫺2.0 0.0 5.0 0.0 9.0 3.0 3.0 2.0 ⫺5.0
⫺4.9 7.6 0.7 ⫺1.4 2.4 4.2 1.2 0.0 0.0 0.0 0.0 0.0 0.0 ⫺11 0.0 0.0
7.0 12.5 1.2 6.1 12.2 5.7 4.9 3.6 0.0 5.0 4.0 21.0 3.0 17.3 2.0 12.1
the ability to quantify the patient’s alignment in the six degrees of freedom. This method provides easy and precise feedback regarding the patient position based on the 3D image localization, very similar to optical systems already used for intracranial targets. The tracking accuracy of our 3D ultrasound imageguided system was measured to be better than 1.5 mm at various target depths, therefore demonstrating the potential of this technique for precision patient positioning. It is important to note that these measurements represent the overall accuracy in a best-case scenario, and integrate all the steps required for its use in a radiotherapy environment (resolution of CT images, ultrasound tracking accuracy, and optical tracking of CT volume for repositioning). Additional inaccuracy may result from user interpretation and registration of the ultrasound images. In this study, all image registrations were performed by a single physician, and repeat ultrasound volumes were acquired to confirm the
Volume 55, Number 4, 2003
accuracy of the initial registration. Variability in image interpretation may exist depending on both the skill of the operator and also the visibility of the specific tumor on ultrasound images. Studies are ongoing to assess such variability as a function of user and tumor location. Our clinical observations with the first patients treated with extracranial radiosurgery reported here demonstrate the feasibility and acute safety of this approach. On average, the ultrasound-based localization required an average shift of approximately 7 mm from the planned position. Although this shift seems relatively large, it is consistent with other extracranial radiosurgery data indicating shifts of 5– 8 mm based on repeat CT scanning (12). Furthermore, shifts were confirmed using repeat ultrasound imaging and registration. When such large PTV misalignments are corrected, however, one must be concerned that the high-dose regions may move closer to organs at risk, hence decreasing the amount of normal tissue sparing. This should be monitored clinically, but can only be adequately addressed with future integration of treatment planning systems with imageguided delivery systems. Other future improvements of the system include the ability to gate the radiation beam based on internal target motion as perceived from real-time ultrasound imaging. Considering the currently available technology, the degree of accuracy achieved using this technique is acceptable in properly selected patients and provides a noninvasive treatment alternative with a high level of patient satisfaction. Further investigation is required, but this technique, using a single, high-dose treatment, offers the potential for rapid, effective palliation with low morbidity. The application of this system has potential not only in treating previously irradiated volumes but also as a boost therapy to areas receiving conventional radiation therapy. Eventual optimal clinical application of extracranial radiosurgical treatment will undoubtedly, like intracranial radiosurgery, not only address accuracy but also require significant attention to patient selection, long-term patient follow-up, and outcome.
REFERENCES 1. Bajada C, Selch M, De Salles A, et al. Application of stereotactic radiosurgery to the head and neck region. Acta Neurochir Suppl 1994;62:114 –117. 2. Buatti JM, Friedman WA, Bova FJ, Mendenhall WM. Linac radiosurgery for locally recurrent nasopharyngeal carcinoma: Rationale and technique. Head Neck 1995;17:14 –19. 3. Chua DT, Sham JS, Hung KN, et al. Stereotactic radiosurgery as a salvage treatment for locally persistent and recurrent nasopharyngeal carcinoma. Head Neck 1999;21:620 – 626. 4. Cmelak AJ, Cox RS, Adler JR. Radiosurgery for skull base malignancies and nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 1997;37:997–1003. 5. Firlik KS, Kondziolka D, Lunsford LD, et al. Radiosurgery for recurrent cranial base cancer arising from the head and neck. Head Neck 1996;18:160 –165; discussion 166. 6. Kaplan ID, Adler JR, Hicks WL Jr, et al. Radiosurgery for
7.
8.
9.
10.
11.
palliation of base of skull recurrences from head and neck cancers. Cancer 1992;70:1980 –1984. Kocher M, Voges J, Staar S, et al. Linear accelerator radiosurgery for recurrent malignant tumors of the skull base. Am J Clin Oncol 1998;21:18 –22. Kondziolka D, Lunsford LD. Stereotactic radiosurgery for squamous cell carcinoma of the nasopharynx. Laryngoscope 1991;101:519 –522. Samblas JM, Bustos JC, Gutierrez-Diaz JA, et al. Stereotactic radiosurgery of the foramen magnum region and upper neck lesions: Technique modification. Neurol Res 1994;16:81– 82. Hamilton AJ, Lulu BA. A prototype device for linear accelerator-based extracranial radiosurgery. Acta Neurochir Suppl 1995;63:40 – 43. Hamilton AJ, Lulu BA, Fosmire H, et al. Preliminary clinical experience with linear accelerator-based spinal stereotactic radiosurgery. Neurosurgery 1995;36:311–319.
Extracranial radiosurgery
12. Lax I, Blomgren H, Naslund I, et al. Stereotactic radiotherapy of malignancies in the abdomen: Methodological aspects. Acta Oncol 1994;33:677– 683. 13. Wang T, Solberg T, Medin P, Boone R. Infrared patient positioning for stereotactic radiosurgery of extracranial tumors. Comput Biol Med 2001;31:101–111. 14. Lohr F, Debus J, Frank C, et al. Noninvasive patient fixation for extracranial stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 1999;45:521–527. 15. Takacs I, Hamilton AJ. Extracranial stereotactic radiosurgery: Applications for the spine and beyond. Neurosurg Clin North Am 1999;10:257–270. 16. Takacs II, Hamilton AJ, Lulu B, et al. Frame based stereotactic spinal radiosurgery: Experience from the first 19 patients treated. Stereotact Funct Neurosurg 1999;73:69. 17. Troccaz J, Laieb N, Vassal P, et al. Patient setup optimization for external conformal radiotherapy. J Image Guid Surg 1995; 1:113–120. 18. Adler JR Jr, Chang SD, Murphy MJ, et al. The Cyberknife: A frameless robotic system for radiosurgery. Stereotact Funct Neurosurg 1997;69:124 –128. 19. Lattanzi J, McNeeley S, Hanlon A, et al. Ultrasound-based stereotactic guidance of precision conformal external beam radiation therapy in clinically localized prostate cancer. Urology 2000;55:73–78. 20. Murphy MJ, Adler JR Jr, Bodduluri M, et al. Image-guided radiosurgery for the spine and pancreas. Comput Aided Surg 2000;5:278 –288. 21. Bouchet LG, Meeks SL, Goodchild G, et al. Calibration of
● S. L. MEEKS et al.
22. 23. 24. 25. 26. 27. 28.
29. 30.
1101
three-dimensional ultrasound images for image-guided radiation therapy. Phys Med Biol 2001;46:559 –577. Ryken TC, Meeks SL, Buatti JM, et al. Ultrasonic guidance for spinal extracranial radiosurgery: Technique and application for metastatic spinal lesions. Neurosurg Focus 2001;11:8. Bouchet LG, Meeks SL, Bova FJ, et al. 3D ultrasound image guidance for high precision extracranial radiosurgery and radiotherapy. Radiosurgery 2002;4:262–278. Bova FJ, Buatti JM, Friedman WA, et al. The University of Florida frameless high-precision stereotactic radiotherapy system. Int J Radiat Oncol Biol Phys 1997;38:875– 882. Buatti JM, Bova FJ, Friedman WA, et al. Preliminary experience with frameless stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 1998;42:591–599. Meeks SL, Bova FJ, Wagner TH, et al. Image localization for frameless stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 2000;46:1291–1299. Ryken TC, Meeks SL, Pennington EC, et al. Initial experience with frameless stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 2001;51:1152–1158. Tome WA, Meeks SL, Orton NP, et al. Commissioning and quality assurance of an optically guided 3D ultrasound target localization system for radiotherapy. Med Phys 2002;29:1781–1788. Tome WA, Meeks SL, Buatti JM, et al. A high-precision system for conformal intracranial radiotherapy. Int J Radiat Oncol Biol Phys 2000;47:1137–1143. Meeks SL, Buatti JM, Bova FJ, et al. Treatment planning optimization for linear accelerator radiosurgery. Int J Radiat Oncol Biol Phys 1998;41:183–197.
doi:10.1016/S0360-3016(02)04406-1
PHYSICS CONTRIBUTION
ULTRASOUND-GUIDED EXTRACRANIAL RADIOSURGERY: TECHNIQUE AND APPLICATION SANFORD L. MEEKS, PH.D.,* JOHN M. BUATTI, M.D.,* LIONEL G. BOUCHET, PH.D.,† FRANCIS J. BOVA, PH.D.,† TIMOTHY C. RYKEN, M.D.,‡ EDWARD C. PENNINGTON, M.S.,* KATHLEEN M. ANDERSON, C.M.D.,* AND WILLIAM A. FRIEDMAN, M.D.† Departments of *Radiation Oncology and ‡Neurosurgery, University of Iowa, Iowa City, IA; †Department of Neurological Surgery, University of Florida, Gainesville, FL Purpose: Stereotactic radiosurgery is an effective treatment modality for many intracranial lesions, but target mobility limits its utility for extracranial applications. We have developed a new technique for extracranial radiosurgery based on optically guided three-dimensional ultrasound (3DUS). The 3DUS system provides the ability to image the target volume and critical structures in real time and determine any misregistration of the target volume with the linear accelerator. In this paper, we describe the system and its initial clinical application in the treatment of localized metastatic disease. Methods and Materials: The extracranial stereotactic system consists of an ultrasound unit that is optically tracked and registered with the linear accelerator coordinate system. After an initial patient positioning based on computed tomographic (CT) simulation, stereotactic ultrasound images are acquired and correlated with the CT-based treatment plan to determine any soft-tissue shifts between the time of the planning CT and the actual treatment. Optical tracking is used to correct any patient offsets that are revealed by the real-time imaging. Results: Preclinical testing revealed that the ultrasound-based stereotactic navigation system is accurate to within 1.5 mm in comparison with an absolute coordinate phantom. Between March 2001 and March 2002, the system was used to deliver extracranial radiosurgery to 17 metastatic lesions in 16 patients. Treatments were delivered in 1 or 2 fractions, with an average fractional dose of 16 Gy (range 12.5–24 Gy) delivered to the 80% isodose surface. Before each fraction, the target misalignment from isocenter was determined using the 3DUS system and the misalignments averaged over all patients were anteroposterior ⴝ 4.8 mm, lateral ⴝ 3.6 mm, axial ⴝ 2.1 mm, and average total 3D displacement ⴝ 7.4 mm (range ⴝ 0 –21.0 mm). After correcting patient misalignment, each plan was delivered as planned using 6 –11 noncoplanar fields. No acute complications were reported. Conclusions: A system for high-precision radiosurgical treatment of metastatic tumors has been developed, tested, and applied clinically. Optical tracking of the ultrasound probe provides real-time tracking of the patient anatomy and allows computation of the target displacement before treatment delivery. The patient treatments reported here suggest the feasibility and safety of the technique. © 2003 Elsevier Science Inc. Extracranial stereotactic radiosurgery, 3D ultrasound, Image guidance, Optical tracking, Metastases, Patient positioning.
INTRODUCTION
tached to the ring. The image coordinates are transformed into stereotactic coordinates through knowledge of the fiducial geometry, and hence the stereotactic coordinates of the target volume are determined. Radiation treatment plans are generated based on the three-dimensional (3D) digital image data, and are followed by positioning the actual patient in the treatment vault by using the stereotactic coordinates mathematically determined relative to the stereotactic ring. This rigid localization system allows extremely accurate targeting of the radiation beam; a dose distribution can be
Stereotactic radiosurgery was defined by Lars Leksell as single-fraction irradiation of intracranial targets that could replace open surgery for selected patients. Historically, this procedure has been performed using an invasive stereotactic head frame. The frame is attached to the patient under local anesthesia using four aluminum pins. This ring system provides a rigid frame of reference for all imaging and target localization. Subsequent to head ring placement, images are acquired using a system of fiducials that are at-
Acknowledgment—Development of the ultrasound navigational system was performed in collaboration with ZMed, Inc., Ashland, MA. Received Jul 19, 2002, and in revised form Oct 31, 2002. Accepted for publication Nov 11, 2002.
Reprint requests to: Sanford L. Meeks, University of Iowa, Department of Radiation Oncology, W189Z-GH, 200 Hawkins Dr., Iowa City, IA 52242-1077. Tel: (319) 356-0881; Fax: (319) 356-1530; E-mail: [email protected] Funding for this work was provided by Whitaker Foundation Biomedical Engineering Research Grant RG-98-0278. 1092
Extracranial radiosurgery
generated that conforms very tightly to the target volume while delivering a very small fraction of that dose to the normal tissues just a few millimeters away. To accomplish this steep dose gradient, multiple, noncoplanar beams of radiation are stereotactically directed on the target. Advantages of this minimally invasive form of treatment for brain tumors are that it is a single outpatient treatment, and that radiation exposure to normal brain tissue is reduced and hence many of the complications associated with conventional radiation therapy are avoided. Although the rigid stereotactic fixation enables accurate radiation treatment, the head ring has limited the procedure to the treatment of intracranial lesions. Because of the success of stereotactic radiosurgery for many intracranial applications, there has been an interest in extending the technology to other treatment sites. For example, several groups have used standard head ring localization to treat primary and recurrent head-and-neck cancers that are confined to the base of skull and upper cervical spine (1–9). Several methods have been proposed to further extend stereotactic localization to extracranial radiosurgery (10 –16). High-precision extracranial radiation delivery is challenging, however, because the target position can shift relative to bony anatomy between the time of image acquisition and the time of treatment. Therefore, several groups are actively investigating real-time imaging at the time of treatment as a mechanism for improved high-precision radiotherapy localization (17–20). Our group has recently investigated the use of optically guided 3D ultrasound for high-precision radiation delivery (21–23). The purpose of this paper is to report on the initial clinical use of the system for radiosurgery of metastatic extracranial lesions. METHODS AND MATERIALS Optical guidance Three-dimensional treatment planning in radiation therapy requires acquisition of 3D image sets, such as computed tomography (CT), magnetic resonance imaging, or positron emission tomography, relative to some fiducial system. A virtual simulation of the patient’s treatment is then performed using these 3D image data. In other words, a model of the patient is created in a computerized treatment planning system using the imaging data. The placement of radiation beams is modeled on the virtual patient, and radiation dose distributions are calculated and displayed on the images. Radiosurgery is unique in that it relies on 3D, or stereotactic, image localization with a robust fiducial system, thereby enabling co-identification of the virtual target in the treatment-planning computer with the actual target position for treatment delivery. To help minimize errors, traditional radiosurgery uses a minimally invasive head ring to coordinate the virtual and real worlds and to minimize motion during image acquisition and treatment delivery. We have previously described a system for noninvasive localization of intracranial targets based on optical tracking of a dental tray attached to the patient’s maxillary dentition
● S. L. MEEKS et al.
1093
(24 –26). Optical tracking is simply a means of determining in real time the position of an object by tracking the positions of either active or passive infrared markers attached to the object. The position of the point of reflection is determined using a camera system. In the commercial version of this system (RadioCameras, ZMed, Inc., Ashland, MA), patient localization is accomplished through detection of four passive markers attached to a custom bite plate that links to the maxillary dentition of the patient to form a rigid system (Fig. 1a). This tracking is accomplished in real time using a charge-coupled device (CCD) optical system that is equipped with infrared illuminators that flood the room with infrared light. The infrared light is reflected off of the four passive markers, and these reflections are read using the CCD optics. The position of the markers is then input to a computer system that combines this information with the stereotactic CT localization of the patient (based on this same dental tray fiducial system) and a calibration matrix that relates the camera position to the linac isocenter. The output from the computer is the patient’s displacement from the desired position (i.e., target at linac isocenter), thereby tracking the patient relative to isocenter (Fig. 1b). The ability of the tracking to occur continuously with updated localization approximately 15 times per second makes realtime positioning and patient monitoring practical. Although this system was originally designed for fractionated stereotactic radiotherapy, it has proven to provide extremely accurate localization relative to the treatment room, and has therefore been used for single-fraction radiosurgery in a select cohort of patients (27). Optically guided ultrasound for extracranial radiosurgery System description. While the initial system for opticalguided radiotherapy provides a submillimeter localization accuracy (25–27), it has been limited to intracranial therapy. Outside of the cranium, soft-tissue targets can move relative to rigid fixation points (e.g., bone structures) between the times of image acquisition, treatment planning, and treatment delivery. Therefore, real-time imaging is required to establish extracranial stereotactic localization of the lesion at the time of treatment delivery. We have developed a system for 3D ultrasound guidance (commercial version SonArray, ZMed, Inc., Ashland, MA) to correct for these misalignments at the time of treatment. Ultrasound was chosen because it is an inexpensive, yet flexible and highresolution imaging modality that can easily be adapted for use in a radiation therapy treatment room. The interpretation of two-dimensional ultrasound images is difficult, however, and can be highly dependent on the skill and expertise of the operator in manipulating the transducer and mentally transforming the 2D images into a 3D tissue structure. Much of this difficulty results from using a spatially flexible 2D imaging technique to view 3D anatomy. Three-dimensional ultrasound reconstruction helps overcome this limitation. We generate 3D ultrasound data sets through optical tracking of free-hand acquired 2D ultrasound data. The operator holds the ultrasound probe and manipulates it over the
1094
I. J. Radiation Oncology
● Biology ● Physics
Volume 55, Number 4, 2003
Fig. 1. (a) The intracranial optical-guided radiotherapy system relies on an array of four passive markers attached to the maxillary dentition using a custom bite plate. Infrared reflections off of the passive markers are read by CCD optics, then input to a computer system which then determines the patient’s position based on these reflections. (b) The computer monitor outputs information about the patient’s displacement from isocenter. The top three numbers indicate the patient’s translational misalignments in millimeters along three principal axes, whereas the bottom three numbers indicate the rotational misalignments in degrees about each of these three axes. The middle number, labeled “vector,” is the root mean square of the three translational misalignments, and is therefore equal to the 3D error. The numbers update in real time, allowing the treatment technologist to correct the misalignments.
anatomic region of interest. The raw 2D data are transferred to a computer workstation using a standard video link. The position and angulation of the ultrasound probe in any arbitrary orientation is determined using an array of four
infrared light-emitting diodes (IRLEDs) attached to the probe (Fig. 2a). Similar to our system for intracranial optic guidance, CCD cameras are used to determine the positions of the IRLEDs, and this information is input to the computer
Extracranial radiosurgery
● S. L. MEEKS et al.
1095
Fig. 2. (a) The ultrasound probe is tracked via an array of four IRLEDs rigidly attached to it. A rocking cradle is used to help guide the free-hand acquisition and to ensure that no gaps exist in the final 3D volume. (b) By tracking the probe, an ultrasound volume can be generated. Shown are three orthogonal views of a transabdominal prostate scan.
workstation. The position of each ultrasound pixel can therefore be determined using the IRLEDs, and an ultrasound volume can be reconstructed by coupling the position information with the raw ultrasound data (Fig. 2b). In addition to building the 3D image volume, image
optical guidance is used to determine the absolute position of the ultrasound volume in the treatment room coordinate system. Because the relative positions of the ultrasound volume and the ultrasound probe are fixed, the knowledge of the probe position in the treatment room coordinate
1096
I. J. Radiation Oncology
● Biology ● Physics
system at the time of data acquisition is sufficient to determine the position of the image volume relative to the linac isocenter. The determination of the relative position of the image and probe corresponds to a calibration step that is performed at the time of system installation (21). Preclinical testing. The accuracy of optically guided 3D ultrasound for extracranial radiosurgery patient positioning was tested using a specially designed absolute ultrasound phantom (21, 23). This phantom consists of 15 echoic spheres imbedded in a tissue-equivalent nonechoic medium. The spheres are arranged in sets of five located at three different depths: 30, 60, and 130 mm. A CT scan (0.51 mm ⫻ 0.51 mm ⫻ 1.25 mm) of the phantom was acquired with a passive infrared tracking array attached to it. An infrared array was used to track the position of the phantom in the room coordinate system. Then, the ultrasound phantom was placed in the treatment vault and the ultrasound probe was fixed on top of the phantom. Coupling of the probe and the phantom was achieved with a thin layer of water. Next, a sphere within the phantom was selected as the target sphere and positioned at the room isocenter based on the CT-optical tracking of the phantom. Next, the localization of the target sphere was determined using our 3D ultrasound guidance system. The 3D ultrasound volume of the sphere was acquired using optical tracking as described above. The 3D ultrasound– based position of the target sphere was determined by finding the center of the sphere in the image using a circle tool placed on each three-orthogonal ultrasound views. The target localization accuracy using the 3D ultrasound optic guided system was thus determined by comparing its experimentally determined position to the position of the target sphere as determined from the CT scan. More complete details regarding preclinical testing, commissioning, and quality assurance of the optically guided ultrasound system can be found in the literature (21, 23, 28). Clinical application. Clinical use of ultrasound image guidance proceeds as follows. Before CT scanning, the patient is immobilized using a custom vacuum cushion as is commonly used in radiation therapy (Vac-Loc, Med-Tec, Inc., Orange City, IA). The CT is acquired with the patient immobilized in the same position that will be used during the radiotherapy treatment to maintain a generally consistent position of mobile anatomy. The CT images are transferred to our 3D treatment planning system (Pinnacle3, ADAC, Milipitas, CA), where the tumor volume and normal structures of interest are delineated. A treatment plan is then designed to conform the prescription dose closely to the planning target volume (PTV), while minimizing the dose to the nearby normal structures. This is achieved by using maximally separated noncoplanar beams. Maximal beam separation provides unique beam entry and exit pathways for each beam, thereby optimizing the dose gradient outside of the target volume. In addition, maximal beam separation provides the most unique beam’s-eye views (BEV) of the target volume, thereby maximizing the plan conformality. This is an extension of geometric planning
Volume 55, Number 4, 2003
Fig. 3. Diagrammatic representation of the beam geometries available for treatment planning.
techniques used in intracranial stereotactic therapy (29), with the primary difference being that intracranial radiotherapy planning allows nearly unrestricted beam entrance over 2 space (i.e., the top half of the patient’s head). Due to possible collisions of the gantry with the patient or the treatment couch, the available geometry for beam entrance is much more limited in extracranial planning. The generally available geometry is illustrated in Fig. 3. When the treatment couch is at its home position (0°, International Electrotechnical Commission angle convention), the gantry is capable of full rotation about the patient. As the couch is moved away from its home position, the gantry is capable of rotating approximately ⫾30° from vertical, creating a cone of possible beam entrances above the patient. As a starting point, the gantry and treatment table angles are chosen to achieve maximal beam separation within this limited space of achievable treatment geometries. The planner then inspects the beam’s eye views for each of these beams and modifies the gantry and table angles to avoid critical structures or minimize the projection of the PTV. Each beam is shaped to match the BEV projection of the planning target volume using a multileaf collimator (MLC) that has a 5-mm leaf resolution at isocenter (Millennium MLC-120, Varian Oncology Systems, Palo Alto, CA). The field shapes are designed with zero margin added to the BEV projection of the PTV, and MLC leaves at the edge of the target. The day of the treatment, the patient is placed in the same immobilization cushion that was used during CT scanning. The patient is initially set up relative to isocenter using conventional laser alignment. A 3D ultrasound volume is then acquired and reconstructed in the computer workstation. The target volume and critical structure outlines, as delineated on the planning CT scans, are overlaid on the acquired ultrasound volume in relation to isocenter. The
Extracranial radiosurgery
contours determined from the CT scans are then manipulated until they align with the anatomic structures on the ultrasound images. The amount of movement required to align the contours with the ultrasound images determines the magnitude of the target misregistration with isocenter based on conventional setup techniques. The target is then placed at the isocenter by tracking an infrared array attached to the treatment couch, which allows precise translation from the initial position to the 3D ultrasound determined position. Once all of the setup information has been verified using repeat ultrasound acquisition and coregistration, treatment proceeds as planned. Initial clinical experience Between March 2001 and March 2002, 16 patients with 17 extracranial metastatic lesions have been treated with ultrasound-guided stereotactic radiosurgery at the University of Iowa. Institutional Review Board approval was obtained for review of these patient treatments. Patient characteristics are shown in Table 1. The patients ranged in age from 51 to 79, with a mean age of 64. There were 10 males and 6 females. Treatment sites were liver/porta hepatis (4), paravertebral/t-spine (3), sacral (2), low neck (2), chest wall (2), hip (1), iliopsoas muscle (1), and arm (1). Eleven of the 16 patients had prior conventional irradiation to the treatment area either for treatment of primary disease or initial presentation of metastatic disease. RESULTS Absolute accuracy of 3D ultrasound image guidance As described earlier, the room positions of the 15 spheres in the absolute phantom were determined experimentally using the 3D ultrasound system, and compared against their known positions. Averaging the difference between the measured and expected for all 15 spheres at all depths, the
● S. L. MEEKS et al.
1097
Table 2. Accuracy of image-guided 3D ultrasound as a function of the depth of the target Depth (mm)
Anteroposterior distance (mm)
Lateral distance (mm)
Axial distance (mm)
30 60 130 All depths
0.4 ⫾ 0.2 0.1 ⫾ 0.1 0.6 ⫾ 0.1 0.3 ⫾ 0.4
0.6 ⫾ 0.1 0.7 ⫾ 0.2 1.4 ⫾ 0.4 0.9 ⫾ 0.4
0.2 ⫾ 0.3 0.5 ⫾ 0.3 0.3 ⫾ 0.4 0.3 ⫾ 0.2
full accuracy of this system for patient positioning was found to be on average anteroposterior (AP) ⫽ 0.3 mm, lateral ⫽ 0.9 mm, axial ⫽ 0.3 mm. More detailed analysis can be found in Table 2, but over all depths the accuracy of target repositioning using our ultrasound image-guided system was maintained below 1.5 mm, and generally below 1.0 mm. Initial clinical experience Treatment planning data for the 16 initial patients can be found in Table 3. The average lesion volume was 75.6 cc (range 3.9 –197.3 cc). A conformal treatment plan was generated for each patient; the average number of beams used was 9.1, with a range from 6 to 11 noncoplanar beams. A typical extracranial radiosurgery treatment plan is shown in Fig. 4, illustrating the conformality and dose gradients achievable using this planning methodology. The prescription isodose volume (i.e., the dose prescribed to the lesion periphery) was chosen such that ⱖ95% of the PTV was inside of the prescription isodose volume. The average prescription isodose to target volume ratio (PITV) was 1.75 (range 1.17–2.69). Fourteen of the 16 patients received single-fraction treatment; for these patients, the mean treatment dose to the lesion periphery was 16.4 Gy, with a range of 12.5–24 Gy. The two remaining patients each received
Table 1. Characteristics of patients treated with extracranial radiosurgery Patient No.
Metastatic lesion location
Primary tumor
Age
Previous dose to treatment region (Gy)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Low neck Lung (chest wall) Hip Liver (2 lesions) Paravertebral (T12-L1) T spine Liver Iliopsoas muscle Porta hepatis Neck Sacral Liver Sacral Arm Liver/Rib T spine
Neck Lung Lung Breast Lung Esophagus Colorectal Spindle cell Ovarian Anaplastic thyroid Renal cell Breast Hepatocellular SCCA skin Hepatocellular Esophagus
79 68 51 57 55 65 76 79 51 67 70 58 65 58 53 66
50 60 30 No 66.6 50.4 No No No 30 42.5 No 30 30 30 50.4
SCCA ⫽ squamous cell carcinoma.
1098
I. J. Radiation Oncology
● Biology ● Physics
Volume 55, Number 4, 2003
Table 3. Characteristics of extracranial radiosurgery treatment plans Patient No.
Volume (cc)
Fields
Dose/fraction (Gy)
Fractions
Isodose line (%)
PITV
1 2 3 4
77.2 51.1 19.6 82.4 19.9 81.2 26.9 125.3 9.6 156.6 81.5 197.3 3.9 20.9 83.2 161.3 88.0
7 11 11 11 11 9 7 9 11 9 6 7 11 9 7 9 10
15.0 17.5 17.5 15.0 15.0 15.0 12.5 12.5 24.0 12.5 12.5 17.5 20.0 17.5 17.5 17.5 12.5
1 1 1 1 1 1 1 2 1 2 1 1 1 1 1 1 1
80 80 80 80 80 80 80 90 80 80 80 80 80 80 80 80 70
1.17 1.36 1.60 1.43 1.54 1.53 1.63 1.65 2.00 1.79 1.71 1.63 2.69 2.06 1.62 1.72 2.60
5 6 7 8 9 10 11 12 13 14 15 16
PITV ⫽ prescription isodose to target volume ratio.
25.0 Gy in 2 fractions of 12.5 Gy each. The average prescription isodose surface was 80% (range 70 –90%), normalized to the point of maximum dose. Although this choice of dose prescription results in a significant dose inhomogeneity when compared with conventional radiotherapy, the 80% prescription isodose was chosen for consistency with intracranial radiosurgery (30). The images and treatment plan information were then transferred to the SonArray treatment guidance station. The patient was placed on the treatment couch, and was initially positioned using conventional laser positioning. An optically guided 3D ultrasound volume was acquired, and the contours of the tumor volume and critical structures, as determined in the planning CT scan, were overlaid on the ultrasound volume. The outlines were manipulated until the best fit between the contours and the ultrasound volume was obtained, as illustrated in Fig. 5. The target misalignments from isocenter averaged over all patients were AP ⫽ 4.8 mm, lateral ⫽ 3.6 mm, axial ⫽ 2.1 mm, and average total 3D displacement ⫽ 7.4 mm (range ⫽ 0 –21.0 mm). Data for
all patients are shown in Table 4. Using optical guidance, the patient misalignment was corrected. A repeat ultrasound acquisition and registration was performed to confirm any patient shift, and treatment was accomplished as planned. No acute complications were reported. DISCUSSION Radiosurgery has dramatically changed the treatment for many central nervous tumors including metastatic lesions, but the need for high-precision treatment delivery has limited its utility primarily to intracranial applications. Extracranial application of high-precision radiosurgery is more complex because motion of soft-tissue targets relative to bony references between the time of image acquisition and treatment delivery is unavoidable. We have developed a system that relies on 3D ultrasound optical guidance to precisely correct for internal organ motion, and adjust the position of the target at the isocenter of the treatment. Because our image-guided system uses 3D images, it has
Fig. 4. The treatment isodose distribution is shown superimposed on axial, sagittal, and coronal CT images through the center of the volume. The center isodose line surrounding the PTV is the treatment line (80%), with the 40% and 16% lines also shown.
Extracranial radiosurgery
● S. L. MEEKS et al.
Fig. 5. The basic screen layout for SonArray shows three orthogonal views through the ultrasound volume with the CT contours superimposed. (a) SonArray layout for targeting an iliopsoas metastasis. (b) SonArray layout for targeting a sacral metastasis.
1099
1100
I. J. Radiation Oncology
● Biology ● Physics
Table 4. Misalignment of target volume from isocenter, as measured using optically guided 3D ulstrasound Shift (mm) Patient No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
AP
Lateral
Axial
3D (root mean square)
4.0 ⫺8.2 0.8 ⫺3.6 3.0 ⫺3.9 3.7 3.0 0.0 0.0 ⫺4.0 19.0 0.0 ⫺13 0.0 11.0
3.0 ⫺5.6 0.5 4.7 ⫺11.6 0.0 ⫺3.0 ⫺2.0 0.0 5.0 0.0 9.0 3.0 3.0 2.0 ⫺5.0
⫺4.9 7.6 0.7 ⫺1.4 2.4 4.2 1.2 0.0 0.0 0.0 0.0 0.0 0.0 ⫺11 0.0 0.0
7.0 12.5 1.2 6.1 12.2 5.7 4.9 3.6 0.0 5.0 4.0 21.0 3.0 17.3 2.0 12.1
the ability to quantify the patient’s alignment in the six degrees of freedom. This method provides easy and precise feedback regarding the patient position based on the 3D image localization, very similar to optical systems already used for intracranial targets. The tracking accuracy of our 3D ultrasound imageguided system was measured to be better than 1.5 mm at various target depths, therefore demonstrating the potential of this technique for precision patient positioning. It is important to note that these measurements represent the overall accuracy in a best-case scenario, and integrate all the steps required for its use in a radiotherapy environment (resolution of CT images, ultrasound tracking accuracy, and optical tracking of CT volume for repositioning). Additional inaccuracy may result from user interpretation and registration of the ultrasound images. In this study, all image registrations were performed by a single physician, and repeat ultrasound volumes were acquired to confirm the
Volume 55, Number 4, 2003
accuracy of the initial registration. Variability in image interpretation may exist depending on both the skill of the operator and also the visibility of the specific tumor on ultrasound images. Studies are ongoing to assess such variability as a function of user and tumor location. Our clinical observations with the first patients treated with extracranial radiosurgery reported here demonstrate the feasibility and acute safety of this approach. On average, the ultrasound-based localization required an average shift of approximately 7 mm from the planned position. Although this shift seems relatively large, it is consistent with other extracranial radiosurgery data indicating shifts of 5– 8 mm based on repeat CT scanning (12). Furthermore, shifts were confirmed using repeat ultrasound imaging and registration. When such large PTV misalignments are corrected, however, one must be concerned that the high-dose regions may move closer to organs at risk, hence decreasing the amount of normal tissue sparing. This should be monitored clinically, but can only be adequately addressed with future integration of treatment planning systems with imageguided delivery systems. Other future improvements of the system include the ability to gate the radiation beam based on internal target motion as perceived from real-time ultrasound imaging. Considering the currently available technology, the degree of accuracy achieved using this technique is acceptable in properly selected patients and provides a noninvasive treatment alternative with a high level of patient satisfaction. Further investigation is required, but this technique, using a single, high-dose treatment, offers the potential for rapid, effective palliation with low morbidity. The application of this system has potential not only in treating previously irradiated volumes but also as a boost therapy to areas receiving conventional radiation therapy. Eventual optimal clinical application of extracranial radiosurgical treatment will undoubtedly, like intracranial radiosurgery, not only address accuracy but also require significant attention to patient selection, long-term patient follow-up, and outcome.
REFERENCES 1. Bajada C, Selch M, De Salles A, et al. Application of stereotactic radiosurgery to the head and neck region. Acta Neurochir Suppl 1994;62:114 –117. 2. Buatti JM, Friedman WA, Bova FJ, Mendenhall WM. Linac radiosurgery for locally recurrent nasopharyngeal carcinoma: Rationale and technique. Head Neck 1995;17:14 –19. 3. Chua DT, Sham JS, Hung KN, et al. Stereotactic radiosurgery as a salvage treatment for locally persistent and recurrent nasopharyngeal carcinoma. Head Neck 1999;21:620 – 626. 4. Cmelak AJ, Cox RS, Adler JR. Radiosurgery for skull base malignancies and nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 1997;37:997–1003. 5. Firlik KS, Kondziolka D, Lunsford LD, et al. Radiosurgery for recurrent cranial base cancer arising from the head and neck. Head Neck 1996;18:160 –165; discussion 166. 6. Kaplan ID, Adler JR, Hicks WL Jr, et al. Radiosurgery for
7.
8.
9.
10.
11.
palliation of base of skull recurrences from head and neck cancers. Cancer 1992;70:1980 –1984. Kocher M, Voges J, Staar S, et al. Linear accelerator radiosurgery for recurrent malignant tumors of the skull base. Am J Clin Oncol 1998;21:18 –22. Kondziolka D, Lunsford LD. Stereotactic radiosurgery for squamous cell carcinoma of the nasopharynx. Laryngoscope 1991;101:519 –522. Samblas JM, Bustos JC, Gutierrez-Diaz JA, et al. Stereotactic radiosurgery of the foramen magnum region and upper neck lesions: Technique modification. Neurol Res 1994;16:81– 82. Hamilton AJ, Lulu BA. A prototype device for linear accelerator-based extracranial radiosurgery. Acta Neurochir Suppl 1995;63:40 – 43. Hamilton AJ, Lulu BA, Fosmire H, et al. Preliminary clinical experience with linear accelerator-based spinal stereotactic radiosurgery. Neurosurgery 1995;36:311–319.
Extracranial radiosurgery
12. Lax I, Blomgren H, Naslund I, et al. Stereotactic radiotherapy of malignancies in the abdomen: Methodological aspects. Acta Oncol 1994;33:677– 683. 13. Wang T, Solberg T, Medin P, Boone R. Infrared patient positioning for stereotactic radiosurgery of extracranial tumors. Comput Biol Med 2001;31:101–111. 14. Lohr F, Debus J, Frank C, et al. Noninvasive patient fixation for extracranial stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 1999;45:521–527. 15. Takacs I, Hamilton AJ. Extracranial stereotactic radiosurgery: Applications for the spine and beyond. Neurosurg Clin North Am 1999;10:257–270. 16. Takacs II, Hamilton AJ, Lulu B, et al. Frame based stereotactic spinal radiosurgery: Experience from the first 19 patients treated. Stereotact Funct Neurosurg 1999;73:69. 17. Troccaz J, Laieb N, Vassal P, et al. Patient setup optimization for external conformal radiotherapy. J Image Guid Surg 1995; 1:113–120. 18. Adler JR Jr, Chang SD, Murphy MJ, et al. The Cyberknife: A frameless robotic system for radiosurgery. Stereotact Funct Neurosurg 1997;69:124 –128. 19. Lattanzi J, McNeeley S, Hanlon A, et al. Ultrasound-based stereotactic guidance of precision conformal external beam radiation therapy in clinically localized prostate cancer. Urology 2000;55:73–78. 20. Murphy MJ, Adler JR Jr, Bodduluri M, et al. Image-guided radiosurgery for the spine and pancreas. Comput Aided Surg 2000;5:278 –288. 21. Bouchet LG, Meeks SL, Goodchild G, et al. Calibration of
● S. L. MEEKS et al.
22. 23. 24. 25. 26. 27. 28.
29. 30.
1101
three-dimensional ultrasound images for image-guided radiation therapy. Phys Med Biol 2001;46:559 –577. Ryken TC, Meeks SL, Buatti JM, et al. Ultrasonic guidance for spinal extracranial radiosurgery: Technique and application for metastatic spinal lesions. Neurosurg Focus 2001;11:8. Bouchet LG, Meeks SL, Bova FJ, et al. 3D ultrasound image guidance for high precision extracranial radiosurgery and radiotherapy. Radiosurgery 2002;4:262–278. Bova FJ, Buatti JM, Friedman WA, et al. The University of Florida frameless high-precision stereotactic radiotherapy system. Int J Radiat Oncol Biol Phys 1997;38:875– 882. Buatti JM, Bova FJ, Friedman WA, et al. Preliminary experience with frameless stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 1998;42:591–599. Meeks SL, Bova FJ, Wagner TH, et al. Image localization for frameless stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 2000;46:1291–1299. Ryken TC, Meeks SL, Pennington EC, et al. Initial experience with frameless stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 2001;51:1152–1158. Tome WA, Meeks SL, Orton NP, et al. Commissioning and quality assurance of an optically guided 3D ultrasound target localization system for radiotherapy. Med Phys 2002;29:1781–1788. Tome WA, Meeks SL, Buatti JM, et al. A high-precision system for conformal intracranial radiotherapy. Int J Radiat Oncol Biol Phys 2000;47:1137–1143. Meeks SL, Buatti JM, Bova FJ, et al. Treatment planning optimization for linear accelerator radiosurgery. Int J Radiat Oncol Biol Phys 1998;41:183–197.