The present invention relates to a method for delivering radiation to a target at variable source-to-target distances. In other terms, the invention relates to a method for delivering radiation to a virtual isocenter. The invention has particular application to treatments for patients of larger girth.
Patient positioning systems are used for accurate and reproducible positioning of a patient for radiation therapy, diagnostic imaging, and certain surgical procedures. Immobilization devices support the patient and facilitate precise and accurate guidance for stereotactic interventions for defined three-dimensional target tissue within the patient's body, including the neck, chest, abdomen, pelvis and proximal thighs.
In a typical radiotherapy procedure, a gantry G (
The need for effective patient immobilization for radiation therapy is well documented. Immobilization reduces normal tissue complication rates and allows increased irradiation of the target tissue. Historically, skin marks have been used to aid in target localization and repositioning. However, skin marks may migrate in successive treatments and the markings can shift with respect to underlying deeper target tissues. As a consequence, fiducial markings have been placed on patient immobilization frames, since these markings do not smear, fade or migrate. In some procedures, fiducial markings may be matched to skin markings to properly locate and position the target tissue relative to the isocenter.
To achieve comfortable immobilization, stereotactic body frames have been developed that support the patient on the couch or table T. One such frame F, depicted in
All accelerators are built so that the source of radiation can be rotated around the isocenter that is fixed in space. As explained above, in the typical procedure, the target tissue of the patient is positioned at the isocenter so the tissue can be irradiated from all directions as the accelerator gantry (such as gantry G in
In a typical installation, the position of the radiation source is constrained to a vertical plane and is located at 100 cm from the machine isocenter I. For machines that contain a multi-leaf collimator apparatus, the distance between the machine head and the machine isocenter I may be less than 43 cm. For some large-sized patients and for treatment techniques that require rotation of the patient couch (i.e., non-coplanar, stereotactic body radiation therapy (SBRT)), this distance is not sufficient because of the risk of collisions between the gantry head and either the patient, the couch, or treatment accessories.
This problem has been partially addressed by modified techniques called “extended distance” treatments or “extended distance source-to-axis distance” (EDSAD) treatments that allow a chosen constant distance D between the source and the target tissue, where D is greater than 100 cm (e.g., 120 cm). With these techniques, the center of the target tissue is positioned on the path of beams but not at the machine isocenter I. In order to irradiate the target tissue from different angles using the existing linear accelerators, EDSAD treatment can only be realized through movement of the target tissue center to various points on the surface of a sphere of radius D-100 cm centered at the isocenter of the machine. This set of points on the circle is referred to as a “virtual isocenter” (VIC). From the point of view of the target center of reference, this treatment is also iso-centric, with different radiation beams from the source on the gantry traveling to various points on the sphere of radius D-100 centered on the target tissue center. In other words, the virtual isocenter treatment establishes that for given gantry and patient couch angles, the central ray of the beam starts from the source, passes through the machine's isocenter at 100 cm from the source, and intersects the sphere of radius D-100 cm centered at the isocenter of the machine.
It can be appreciated that in a typical EDSAD treatment, the patient couch will necessarily be moved for each subsequent gantry and couch angle to maintain the center of the target tissue at a point on the virtual isocenter (VIC sphere), since the gantry is constrained to rotate about a fixed horizontal axis passing through the machine isocenter I. Thus, when the gantry is at its 0° position directly overhead the target tissue, the couch T must be lowered so that the tissue center is at a point on the VIC sphere. Likewise, when the gantry is at its 180° position, the couch must be raised. When the gantry is at a 90° position, lateral to the couch, the couch T must be moved laterally away from the gantry to position the target tissue at the VIC sphere.
Certain problems exist with this treatment technique. For instance, there is no easy way to use the room lasers to accurately guide the patient's body to the desired treatment location at the distance D. Since the location of the target tissue cannot be determined a prior in the couch coordinate system, it is not practical to determine the couch translation and rotation during the treatment. Moreover, even if the patient can be moved to the desired location by using coordinate transformation method and couch translational movements, a further problem arises due to bending of the couch that naturally occurs when the couch is cantilevered from the base. It can be appreciated that any cantilevered structure deflects downward due to its own weight. Add to that the weight of a patient and support frame mounted on the couch and it can be seen that the amount of deflection becomes non-negligible. For target tissue of relatively small size these deflection errors can mean the difference between properly irradiating the entire target tissue mass or irradiating more surrounding healthy tissue than target itself. When the couch and patient position is fixed during treatment, the target tissue center can be deliberately placed at the machine isocenter so any couch deflection is immaterial. However, in treatments requiring movement of the couch, couch deflection must be addressed at every new position of the couch. These practical problems make precise positioning of patients at extended distance EDSAD treatments wearisome. As a result, the available geometrical freedom of positioning the patient for optimal radiation exposure is significantly restricted in certain radiation therapies, particularly for SBRT treatments.
The present invention provides a novel approach to treatments using conventional accelerators that eliminates the need for laser guidance of the patient's body to a desired location at the extended treatment distance D. The present invention also eliminates the positioning inaccuracies of the target tissue relative to the beam that may be caused by deformation or bending of the patient couch.
For the purposes of promoting an understanding of the principles of the invention, reference will now be made to the embodiments illustrated in the drawings and described in the following written specification. It is understood that no limitation to the scope of the invention is thereby intended. It is further understood that the present invention includes any alterations and modifications to the illustrated embodiments and includes further applications of the principles of the invention as would normally occur to one skilled in the art to which this invention pertains.
The present invention relates to a process of expanding the treatment capability of linear accelerators and ultimately reducing patient treatment time. The invention integrates the linear accelerator gantry and patient couch with patient immobilization device, wall mounted laser system, target tissue locator, and position planning software. The use of target tissue locating and imaging devices, such as portal imaging, optical cameras, fluoroscopy and CT scanning, can enhance the implementation of the invention.
A typical therapeutic treatment system includes a gantry G that carries the linear accelerator. The gantry is configured to rotate in a single vertical plane about an isocenter I, as shown in
The patient couch T can also rotate about a vertical axis that passes through the isocenter of the gantry. The range of couch rotation is restricted to avoid collisions with the gantry. The table can be positioned at specific angle with an accuracy of about one degree. The couch may also provide translational motion in vertical, lateral and longitudinal directions (relative to the isocenter).
In accordance with the present invention, the patient couch establishes a three-dimensional (x,y,z) coordinate system (see
The present invention contemplates an immobilization frame F (
The treatment room is preferably provided with a tissue locator to locate the target tissue within the patient. This can be achieved by imaging and reconstruction of the patient's three-dimensional anatomy. The images include CT data and may be supplemented by MRI or PET images. The three-dimensional images allow the treatment planner to determine the locations and sizes of the target tissue (e.g., tumors), as well as the sensitive strictures surrounding those targets. The three-dimensional images are related to fiducials 12 on the immobilization frame F, so that the target tissue positions are defined in the frame coordinate system. The position of fiducials is fixed relative to the origin of the frame coordinate system (OX, OY, OZ)
Planning software is used to determine the treatment angles in terms of the gantry and the couch directions and spatial movements. In accordance with the present invention, this software will evaluate the planned movements and determine whether collisions may result between the gantry and the patient couch or treatment accessories. The software will also warn the planner of treatment configurations that are unachievable at the time of irradiation.
Accurately positioning the patient in a desired position and orientation is the key for successful radiation therapy. In current treatment facilities, there are usually three measurement reference devices used to find the location of the patient relative to the system coordinates—namely, the wall-mounted laser beams, the field light of the linear accelerator, and readings from mechanical sensors associated with the rotating gantry and translating patient table. However, applicants have discovered that many sources of position errors arise when using the current techniques and position reference devices.
Looking first at the patient couch, the typical sensor readings for couch translation is in 1.0 mm (0.1 cm) increments, so that the maximum accuracy (E) of the couch readings in any translational direction is 0.5 mm. Therefore, the maximum combined inaccuracy (Ex2+Ey2+Ez2)0.5 is 0.87 mm. In the rotational degrees of freedom, both the gantry and the couch have one degree accuracy. The maximum accuracy for setting a target tissue at a specific location according to couch and gantry rotation values is thus limited by the accuracy of the couch and rotation measurements.
In experiments, disagreements may arise between the couch position measurements and the position data generated by the wall lasers. For instance, with the tissue positioned at a distance of 50 cm from the isocenter in one experiment, the measurement generated by projected laser beams was about 2.0 mm apart from the reading from couch position data. It is believed that this position discrepancy is caused by either rotation inaccuracy of the couch or inaccuracy of the laser installation. Further experiments showed that the error was primarily attributable to the laser installation. However, the couch motion in the present invention relies principally on the couch readings and not on position data generated by the wall lasers.
It was noticed that vertical position readings of the couch may also be inaccurate. The typical patient couch includes a bed that extends longitudinally toward the gantry relative to a support base. Thus, the bed is essentially cantilevered relative to the base. In most uses, the patient is situated on the couch bed with his/her head adjacent to the gantry. Thus, the cantilevered portion of the couch carries the majority of the patient's mass. To determine the effect of the patient weight to the vertical position reading error, we conducted experiments in that different cantilevered weights were supported on the couch extended to different longitudinal positions. The results of these experiments, summarized in the graph and Table A of
For the purposes of the present invention, the absolute magnitude of the vertical deflections is not as important as the relative change between data cells in the table. For instance, the differences between vertically adjacent cells (i.e., constant longitudinal position with increasing weight) correspond to bending error due to changing weight. The differences between horizontally adjacent cells (i.e., constant weight with increasing longitudinal displacement) correspond to bending error introduced by the couch moving in the longitudinal direction.
A similar experiment was constructed to determine errors introduced by moving lateral movement of the couch. The following Table B summarizes the vertical displacement of the couch as a function of lateral displacement and cantilevered weight:
As the above data reveals, couch movement and patient weight affect the vertical displacement of the table, and consequently causing the error in the couch position measurements. In accordance with one aspect of the present invention, this empirical data may be used to generate an error adjustment map due to bending of the couch as a function of longitudinal displacement, lateral displacement and equivalent weight applied to the couch. Preferably, the error adjustment map may be generated using curve-fitting methods to provide an optimized algorithm, equation or table look-up. It is contemplated that this error adjustment map will provide a couch vertical offset value as a function of input values for patient weight, longitudinal couch position and lateral couch position. It is further contemplated that this error adjustment map will be specific to each type of patient couch, due to mechanical differences between couches.
The experiment data shows that the couch vertical offset due to patient weight and couch position is cumulative, meaning that the vertical offset obtained from Table A for longitudinal movement is combined with the vertical offset from Table B for lateral movement. Thus, in one embodiment of the invention, two error adjustment maps are provided—one for longitudinal movement and one for lateral movement. These error adjustment maps are then used as described in more detail herein to accurately position the center of the target tissue at the virtual isocenter VIC.
The present invention contemplates a new process for radiation treatment using a combination of the couch, the immobilization frame and the linear accelerator.
Initiating the process requires the following input:
With respect to the first required input (the location of the frame origin), it is understood that any fixed point on an immobilization frame may be defined as its origin. To reduce the chance of reading mistakes, the origin of the frame coordinate system is situated as shown in
In one approach to locate the frame origin, the couch is translated with the frame locked to the couch until the frame origin (Ox, Oy, Oz) in the couch coordinate system coincides with the gantry isocenter I. The translational couch movements, or move vector, necessary to move the frame origin to the isocenter will then correspond to the actual location of the frame origin relative to the couch origin. However, this approach is not always practical due to the range limitations of the couch movement. Another approach is to lock the frame to the couch and then drive the couch until a particular known reference point, such as a fiducial, in the frame coordinate system (x, y, z) is located at the isocenter. The couch position vector (Px, Py, Pz) in the couch coordinate system may be expressed as (Px, Py, Pz)=(Ox−x, Oy−y, Oz−z). Therefore, the origin of the frame coordinates (Ox, Oy, Oz) in the couch coordinate system can be determined as (Ox, Oy, Oz)=(Px+x, Py+y, Pz+z). This required input must be obtained for each type of the frame at each locking position on the couch, since the frame is at a different location on the couch for each different locking position.
The second input is the target tissue position in the frame coordinate system. With the patient in the immobilization frame F, an imaging scan, such as CT, MRI or PET, is performed. The target tissue, patient markers and fiducials 12 on the frame are ascertained in the images. The position of the target tissue in frame coordinates can be determined based on the relationship of the target tissue center to the fiducials, which have a known location relative to the frame origin and ultimately to the couch coordinate system, as determined in the prior step.
The final inputs are the desired treatment set-up parameters, such as couch angle, gantry angle and extended distance (especially important for large girth patients), from the treatment plan. The objective of the present invention is to easily and accurately determine the couch positions to satisfy the desired treatment, while avoiding or eliminating possible errors from the various sources described above. In other words, the objective is to achieve the couch and gantry movements necessary to follow the treatment plan without error and without collision between couch and gantry. Thus, these inputs need to be adjusted to correct various potential errors noted above.
One embodiment of the present invention contemplates a series of steps to accurately determine the couch positions during the treatment movements. These steps may include:
Steps (1), (2), and (4) are relatively straightforward and have been discussed above. The remaining steps require more detailed explanation and form an important part of the present invention. Step (3) entails partially correcting the couch vertical position error caused by the weight of the patient and the couch on the cantilevered portion of the couch. The effect of these cantilevered weights on the vertical couch coordinate can be measured and corrected using the following sub-steps of Step (3):
The sub-steps of Step (3) can correct the couch bending error caused by the weight on the couch and caused by the longitudinal position of the couch. The end result of the sub-steps of Step (3) is that accounts for tare by shifting the coordinate system origin from which subsequent position determinations are made. In other words, a revised couch position is provided for use in the subsequent Steps (4)-(6).
Step (5) attempts to correct the error introduced by moving the tissue center from the machine isocenter to the virtual isocenter used in the EDSAD treatment protocol described above. Known coordinate transformation techniques may be used to calculate the destination couch values (Cdx, Cdy, Cdz) according to the desired gantry and couch angles obtained in Step (4) and using the adjusted frame origin (Ox′, Oy′, Oz′) obtained in Step (3). Since the transformation equations depend on how the coordinate systems are defined, the sequence of steps defining a preferred coordinate transformation protocol for the coordinate systems described herein is as follows:
Thus, Step (5) provides a couch movement value that can position the tumor at the virtual isocenter for any desired gantry angle and couch angle. These coordinate transformations are represented in
In accordance with the steps outlined above, the couch movements for the treatment protocol are observed in the couch coordinate system. Thus, a vector transformation is necessary into couch coordinates. This transformation for a movement vector V is given by the following:
where D is the extended distance value, Θ is the gantry rotation angle and Φ is the couch rotation angle. Application of this vector transformation places the center of the target tissue at the point O′ or the VIC for the extended distance treatment.
However, due to the motion of the tumor center from machine isocenter to the virtual isocenter, a new vertical couch bending error is introduced. Thus, the final Step (6) of this process is to correct this vertical bending error. Note that the vertical bending error at the isocenter was initially corrected in Step (3), but since the tissue center has been moved from system isocenter I (or origin O in
(D−100)2=(Ox′−VICCx)2+(Oy′−VICcy)2+(Oz′−VICcz)2
The couch can move at most (D-100) centimeters in any direction from the tare point. When the tare is performed at (Ox′, Oy′, Oz′), all errors are eliminated. However, when the couch is actually moved to VICc according to the calculated transformation value, the target tissue center does not remain at the virtual isocenter, as mathematically predicted, due to changes caused by the weight longitudinal and lateral error relationships expressed in Tables A and B. In order to obtain the true couch movement vector it is first necessary to correct the error caused by the longitudinal motion of the couch. The error adjustment maps discussed above may be consulted to estimate the vertical error caused by the longitudinal movement of the couch VICcz-Oz′ and ultimately to generate a corrected virtual isocenter value of VICcy′.
Next the vertical error due to the lateral motion of the couch is corrected. Again, the error adjustment maps discussed above can be used to estimate the error introduced by the lateral change VICcx-Ox′, and to generate an adjusted coordinate ViCcy′. The same process is instituted to correct for the weight of the couch and patient, again using the error adjustment maps.
All of the steps of this preferred embodiment of the invention can be implemented in software. The system operator need only input the required information set forth above and establish the tare for the patient. The software then performs the coordinate transformations described above to transform the operator input couch and gantry moves to couch lateral, vertical and longitudinal moves that compensate for the errors described herein. The calculations necessary to determine the adjusted couch move protocol may be done immediately when the treatment protocol data is entered and before the treatment is commenced. Thus, if an operator enterers a sequence of desired gantry and couch rotations, software implementing the present invention will determine the appropriate couch movements necessary to maintain the center of the target tissue at the machine isocenter I for standard coplanar treatments or at the virtual isocenter for non-coplanar extended distance treatments. This software may communicate the appropriate couch position data to a control system in the base of the patient couch or table T that is responsible for mechanically moving the couch.
It is contemplated that error adjustment maps for correction of lateral and longitudinal movement errors are maintained in a database accessible by the software. The data populating these maps is preferably specific to each patient table T and may be generated by the couch manufacturer or empirically by tests at the treatment facility.
It is further contemplated that the software implementing the steps of the present invention will work through a user interface to permit entry of the Inputs (a)-(c) described above. Thus, the operator can enter the location of the immobilization frame in couch coordinates (Input (a)) and the target tissue center position in frame coordinates (Input (b)) following the procedures outlined above. The Inputs (c) are independent of the couch and its movement errors and are instead determined by the desired radiation treatment protocol.
The present invention contemplates a method that enables fast and precise positioning of a target tissue at varying source to target distances, or at a virtual isocenter. According to the present invention, a tare is obtained at the system isocenter (in global coordinates), so that the bending error introduced by the effective weight of the patient at this point is already corrected. When the couch is moved in lateral and longitudinal directions, the effective weight of the patient that causes couch or table bending also changes. This change of effective weight is difficult to quantify accurately; however for the purposes of the present invention this change can be estimated. This estimated change due to effective weight can be used for adjusting the bending error.
It has been found that the bending error introduced by lateral and longitudinal movements can be treated separately since these errors are generally independent based on empirical data. This independence means that bending errors can be isolated into a bending error maps as a function of weight only, longitudinal position only, and lateral couch position only.
While the invention has been illustrated and described in detail in the drawings and foregoing description, the same should be considered as illustrative and not restrictive in character. It is understood that only the preferred embodiments have been presented and that all changes, modifications and further applications that come within the spirit of the invention are desired to be protected.
This application claims priority to co-pending provisional application No. 60/647,893, entitled “Method for Radiation Therapy Delivery at Varying Source to Target Distances”, filed on Jan. 28, 2005, the disclosure of which is incorporated herein by reference, and to co-pending provisional application No. 60/647,920, entitled “Relocatable Stereotactic Immobilization Apparatus”, filed on Jan. 28, 2005, the disclosure of which is incorporated herein by reference. This application also claims priority to international provisional application No. PCT/US2006/002883, entitled “Method for Radiation Therapy Delivery at Varying Source to Target Distances”, filed on Jan. 27, 2006, the disclosure of which is incorporated herein by reference, and to international provisional application No. PCT/US2006/002912, entitled “Relocatable Stereotactic Immobilization Apparatus”, filed on Jan. 27, 2006, the disclosure of which is incorporated herein by reference.