CyberKnife: Treatment Planning and Delivery Capabilities Donald A. McRae, Ph.D. Department of Radiation Medicine Georgetown University Hospital
CyberKnife System
gun deck X-ray source magnetron Accelerating waveguide 9 GHz cables to modulator in separate room Image detectors
Cone set: 5mm 60 mm diameter beams at 80 cm
Cone Factor Comparisons GUH output factor compare 1.05 1 0.95 0.9 Output Factor 0.85 0.8 0.75 0.7 GUH Miami UCSF Vicenza BRMC 0.65 0.6 0.55 0 10 20 30 40 50 60 collimator mm PTW 1 mm 2 diode adopted as standard
Radiation Characteristics
Beam profiles from which Off-Axis Ratios are obtained OAR 7.5, 15, 60 mm diam. collimators 100 90 80 70 60 50 40 30 20 10 0-5 0-4 0-3 0-2 0-1 0 0 1020304050 r (mm) No flattening filter, open air chambers
~ 120 robot beam delivery points (called nodes) As the treatment progresses, the robot walks through a set of nodes in sequence. This is referred to as a path.
nodes path Beams: 12 possible/node lesion Courtesy J. Rodgers
3D-non non-isocentric: CyberKnife
TREATMENT PLANNING FUSION CAPABILITIES CT/MR CT/PET
CONTOURING Fusions and contouring may be done on any TPS with DICOM export and the CT and contours imported into the CyberKnife The CT anatomy is the basis for treatment planning and all subsequent targeting CavS meningioma
Inverse planning has two steps: Conformal Shape Inverse Planning 1. Beam Geometry Algorithm Create a list of all voxel elements along the boundary of each lesion and use this to create a list of all possible target points on the boundary.. For each of the possible ~1200 beams, the algorithm randomly selects one of the target points to aim at. 2. Optimization of beam weights by inverse planning The Treatment Planning System solves a linear optimization problem to determine beam weights with the following goal:: Minimize the total MU delivered during treatment, subject to minimum and maximum absorbed dose constraints in the lesions and maximum absorbed dose constraints in the critical structures. This optimization is solved with Simplex linear programming. One disadvantage is the inability to assess progress towards a solution as you can in typical IMRT inverse planning.
Beams returned from an optimization run Cav Sinus meningioma
Dose calculation The standard algorithm uses a ray-tracing function based on stored beam TMR data, off-axis ratios (OAR) and an output factor for the collimator(s) CF, At each point, the contribution from each beam is given by the equation: d = (SAD/800) -2 * CF * TMR * OAR * MU, where SAD is the source axis distance, along the central axis of the beam, to the point of the dose calculation, and MU is the number of monitor units for the beam. Contour corrections can be implemented for beams which enter the surface at oblique angles or graze the surface. Tissue density corrections are made by the simple path-length method. CT densities for air and lung are defined for this purpose.
Complex Benign Skull Base Lesion Cavernous sinus meningioma
DVH
Another complex benign lesion Acoustic Neuroma
Results for Complex Benign Skull Base Tumors Complex lesions defined: intricate threedimensional shape and spatial entanglement with one or more critical structures. (N=27) The mean tumor volume was Prescription IDL: 79.5(±6)%; 8.46 cm 3 (range: 0.51 34.3 cm 3 ). Tumor coverage: 95 (±3.7)%;( NCI: 1.7 (±0.3)( [range 1.0-2.3], CI: 1.6 (±0.3)( and HI: 1.26 (±0.1).( NCI = TV*PIV TIV 2 1.0 is ideal in this scheme Neither tumor growth nor adverse effects have been noted in any patients during follow up.
Multiple Lesions
BEAM POINTING Two or more lesions may be treated at the same time
Clival Chordoma/ Graves Ophthalmopathy Deliver: Chordoma: 700cGy x 5 and 30% of that to Graves areas
Clival Chordoma/ Graves Ophthalmopathy
Clival Chordoma/ Graves Ophthalmopathy Chordoma: 700cGy x 5 to 80%IDL, 93% coverage. Deliver 30% to Graves areas
Extracranial: e.g. Spinal lesions
Spinal lesion volumes Volumes (cc) 500 400 300 200 100 0 Spinal Lesion Volumes 111 101 91 81 71 61 51 41 31 21 11 1 Series1 A tremendously diverse collection of tumor sizes, shapes, locations, and juxtapositions to the spinal cord and other critical structures were encountered. Most have had previous external beam. Mean volume 109±131 cc Range 0.53-839 cc, Mean length along the cord: 57 mm (9-141mm); Mean closest distance to cord (not canal): 3.2 mm (0-11mm).
H y p o f r a c t i o n a t i o n Spinal radiosurgery starting point guide-lines for target peripheral doses Dose/fraction (cgy) Fractions Examples of use 800 3 Untreated spine: (met. breast, thyroid, colon, renal, bladder, gross melanoma, squamous, small cell lung) 700 5 Unresected chordoma, chondrosarcoma, osteogenic sarcoma 700 4 Resected (microscopic disease remaining) chordoma, chondrosarcoma, osteogenic sarcoma 700 3 Retreatment of spinal lesions after external beam (met. renal, nsc lung, sc lung, breast, giant cell, melanoma, colon, cervix, adenocarcinoma, bladder) 500 5 Microscopic disease of resistant histology (renal, prostate, melanoma, adenocarcinoma, adenoid cystic) Gross disease of less resistant histology, (leukemia, lymphoma) Benign spinal lesions (neuroma, schwannoma, meningioma, hemangioma, ependymoma)
An example demonstrating a case close to the mean statistics TV = 67 cc; Min. distance to SC= 3mm Rx: 75% IDL 93% coverage, NCI = 1.9 20 mm collimator, 303 beams
Spinal AVM
Lesion Tracking On Treatment
Lesion Tracking On Treatment The angles used for tracking on the CyberKnife are ±45 R and L lateral obliques. Anterior 45 45 Right Left Posterior
CRANIAL TRACKING On treatment, the dynamic tracking system periodically obtains two orthogonal x-ray images of the patient and compares them with DRRs from the treatment planning. This yields the translational and rotational correction values needed to align the two. During treatment these values are transmitted to the robot to perform adaptive targeting to account for these patient position changes.
Fiducial Tracking (a) (b) Examples of fiducials as seen in tracking images: (a) near optimal and (b) among spinal hardware, but still trackable.
Skeletal structure tracking
Early set-up image comparison
Synchrony Respiratory motion tracking First Step: Fiducial Placement
Spiral Breath-hold hold Treatment Planning CT Planning proceeds in the normal way, except that Motion Tracking is selected.
Patient Garment with LED markers
Synchrony Respiratory Tracking System
Synchrony how does it work? Beacons on skin obtain instantaneous respiratory motion. X-ray snapshots of internal position of fiducials are made. A model is built correlating the instantaneous surface position to fiducial (tumor) location. The Robot aims the beam at the tumor at all times based on the surface motion and model.
Synchrony Graphical User Interface
Technology Concept External chest position red light beacons Internal gold fiducials
REAL BREATHING BEING TRACKED
Radiographs Pre and Post Treatment Pre-TX: 1 Week Post Tx 4 Weeks Post Tx
Prostate Treatment: CK boost
IDL (feet first orientation)
Prostate Radiosurgery Tracking
Extremities For s rigid immobilization of the extremity, each patient was fitted with a personalized fiberglass cast. Each cast was bi-valved to allow for easy removal between treatments. This decreased the risk of loss of joint mobility during treatment. s
Left foot of patient with resected synovial sarcoma in bi-valved cast: the five gold seeds inserted into the cast are easily visualized.
Quality Assurance Isocrystal Pointing Accuracy Calibration and Testing
End-to-End Test Alignment of the film with the edges of the cube is crucial
End-to-End Film Analysis Program
End-to-end tests Final results at commissioning March 2002; Targeting error as total displacement: 6-D skull tracking: 0.44 mm,.73 mm for Bicron and CIRS skull tracking phantoms, resp. 6-D fiducial tracking: 0.63 mm CIRS body phantom. Results from June 2002, 3 month QA: Type * Par. Par Orth. Orth. Par Avg. SD (rms) 6-D Skull 0.47 0.72 0.70 1.30 1.06 0.85 0.33 (mm) 6-D Fiducial (mm) 0.83 0.83 0.83
Monthly TLD dose delivery check Adjacent slices showing TLDs in phantom The treatment is set up using CK image tracking and delivered as a patient treatment. Results average within 5% of predicted.
Month/Year Reporting: FY : 7/1/05-thru most rece Day One (3/12/02) thru most recent Synchrony Synchrony Intracranial Sites Treated # of Sites # of Stages cases # of Sites # of Stages cases AVM/AVOM 7 7 Trigeminal Neuralgia 4 4 Acoustic Neuroma 1 5 18 90 Meningioma 5 25 53 254 Pituitary Adenoma 12 58 Pineal Region Tumor 2 8 Glioblastoma GBM 1 1 23 92 Craniopharyngioma Hemangioblastoma 4 12 Chordoma 1 5 6 30 Schwannoma 2 10 Other/Vas/Func Benign Tumors Lung Met 12 13 115 183 Liver Met 2 5 Breast Met 4 6 104 184 Renal Met 5 9 29 46 Colon Met 8 22 Melanoma Met 39 59 Prostate Met 4 7 Pancreas Met 1 5 Ovarian Met 1 5 Other Metastatic Tumor 7 21 45 120 Unknown Metastatic Tumor 6 20 Chondrosarcoma Glomus Tumor Other Primary Tumor 3 15 Astrocytoma/Glioma 1 5 17 82 Ependymoma 4 16 Oligodendroglioma/Medulloblastom 1 5 9 43 Hemangiopericytoma 2 6 Extracranial Sites Treated # of Sites # of Stages Synchrony # of Sites # of Stages Synchrony C-spine 1 5 49 182 T-spine 6 26 100 367 L/S-spine 5 35 85 325 Lung 17 55 17 79 235 48 Liver 16 46 7 Pancreas 3 9 2 26 88 8 Colon 1 5 Bone 4 18 51 190 Head/Neck/ENT 9 37 34 152 Prostate 1 3 4 12 Nasopharynx 4 30 13 99 Other 16 64 1 111 400 2 Total Extracranial Treated 66 282 569 2101 TOTAL TREATED 104 377 20 1089 3484 65
The CyberKnife Team Radiation Oncology Gregory Gagnon, MD Brian Collins, MD K. Wm. Harter, MD Jeff Moulds, MD Anatoly Dritschilo, MD Physics Donald McRae, PhD Sonja Dieterich, PhD Huaying Ji Frank Xia,, PhD Surya Neupane Neurosurgery Therapy Walter Jean, MD Patricia Kornegay,, RTT Fraser Henderson, MD Gerard Elie,, RTT Christopher Kalhorn,, MD Sosena Asrat,, RTT Kevin McGrail,, MD T. Bertram Tucker, RTT