Transcript chamberland
Part I Physicists do it in Hospital Tong Xu Dept. of Physics Carleton University Why there are physicists in the hospital? Medical Physicists Where in the hospital can you find them? Diagnosis imaging departments: • Radiology and Nuclear Medicine (CT, MRI, PET…) Cancer centre • Medical Physics department (Radiotherapy) What is their job? Make sure the equipments are working according to their physics specifications Perform radiotherapy treatment planning Why we need physicists to perform these tasks? Let’s to go back to the history of some of the medical technologies. Related Medical Technologies Three examples … X-ray CT Magnetic Resonance Imaging Radiation Therapy Discovery of X-ray First discovered by German Physicist Wilhelm C. Röntgen in 1895 On a New Kind of Rays Nature 53, 274-276 (23 January 1896) Discovery of X-ray Independently discovered by Nikola Tesla in 1896 Discovery of x-ray 1. Crookes Tube Invented by Sir William Crookes, chemist and physicist, around 1860s. A demonstration of the cathode ray – accelerated electron beam. Discovery of x-ray 2. Cathode ray Cathode ray is a beam of electrons Discovery of x-ray 3. Rontgen’s experiment A mystery radiation was coming out from the tube Röntgen called it X-ray In fact, x-ray is just a ray of light photons with much higher energy than ordinary light Typical x-ray spectrum Medical Application of x-ray Röntgen received the First Physics Nobel price in 1901 X-ray radiograph It’s a shadow image of human What do we need to see through a human? X-ray X-ray Computer tomography X-ray projections of heart CT Image reconstruction Projections at different angle 3D structure http://rpop.iaea.org/ Inventers Theory proposed by a physicist Allan MacLeod Cormack in1956 two papers in the Journal of Applied Physics in 1963 and 1964 First Prototype by electrical engineer Godfrey Hounsfield in 1969 The first CT prototype First Prototype by Godfrey Hounsfield in 1969 Cormack and Hounsfield shared the Medical Nobel prize in 1979 Magnet Resonance Imaging 1. Stern molecule beam (1922) Individual gas molecules fly through a pair of magnets developed by German Physicist Otto Stern and Walther Gerlach in 1922 Magnet Resonance Imaging 2. Some nucleus are like tiny magnets Detector N S N S Magnet Resonance Imaging 2. Some nucleus are like tiny magnets Detector N S N S S N N S Otto Stern received Physics Nobel prize in 1943 Magnet Resonance Imaging 3. Precession of magnetic dipoles Some They nuclear has magnetic momentum are like magnetic dipoles They precess around the external magnetic field Just like a Gyroscope Check out this animation http://www.simplyphysics.com/MRI_shockwave.html B Magnet Resonance Imaging 3. Precession of magnetic dipoles The precession frequency B ω is in the radio frequency range is the Gyromagnetic ratio B Magnet Resonance Imaging 3. Precession of Magnetic dipoles Aligned with the external Magnetic field B Lower energy state B Aligned against the external Magnetic field B Higher energy state The nucleus feel more comfortable to stay in lower energy state Magnet Resonance Imaging 4. Nuclear Magnetic Resonance What if I send nucleurs a Radio wave that has the same frequency as the precession? American physicist Isador I. Rabi had an great idea! Magnet Resonance Imaging 4. Nuclear Magnetic Resonance Detector N S Radio frequency signal ~ N S Magnet Resonance Imaging 4. Nuclear Magnetic Resonance The nucleus will resonance with the RF wave They absorb RF energy And flip to higher energy state Can measure the nuclear magnetic montemtum precisely Isador I. Rabi received Physics Nobel prize in 1944 Magnet Resonance Imaging 5. NMR with solids and liquids In 1946, two other Americans, Edward M. Purcell and the Swiss-born Felix Bloch, separately apply this nuclear magnetic resonance (NMR) method to solids and liquids. Purcell and Bloch received Physics Nobel prize in 1952 Principle of NMR Resonance frequency B Since the gyromagnetic ratio γ is unique for nucleus of each elements Nuclear Magnetic Resonance is a powerful tool for chemical analysis Until 1970s…. Magnet Resonance Imaging 5. Apply NMR to imaging Paul Lauterbur & Peter Mansfield applied NMR to image body in 1970s Introduced Thus, gradients to the magnetic field frequency the radio wave emitted by the nucleus tell us where they are. Magnetic Resonance Imaging MRI scanner Source: sfu.ca MRI A technique for imaging soft tissues source: lecture slides from Prof. I. Cameron Lauterbur and Mansfield received Medical Nobel prize in 2003 Nuclear Medicine x-ray CT Chest X-ray Cancer diagnosis http://www.dcmsonline.org/ Physics in Cancer treatment Radiation Therapy Uses Kills cells ionizing radiation tumour by damaging tumour Radiation therapy External beam radiation therapy Use x-ray generated from linear accelerator. Max energy: 4~20 (MeV, 106 eV) Mega-Electron-Volt Compare to visible light: 2-3 eV Compare to UV light: 3-5 eV 1000,000 times higher than UV light Linear accelerator (Linac) Accelerated high energy electron beam hit a Tungsten target Produce high energy x-ray beam Source: www.cerebromente.org.br Treatment planning It’s also a job for physicists ! X-ray , electrons, photons, scatter radiation dose … Only a medical physicist were trained to deal with them ! Summary Many medical technologies are originated from physics discovery. Then, developed by physicists. Medical physicists are The “customer service” team Improve the techniques Develop new techniques Ionizing radiation Ionizing radiation damages the cell Electron x-ray photons DNA Ionizing radiation X-ray photon DNA Excited by physics discoveries Passionate about People’s well being Positron Emission Tomography (PET) http://www.mni.mcgill.ca/cog/paus/techniques.htm PET image How is x-ray been generated? 1. Bremsstrahlung radiation How is x-ray been generated 2. Characteristic x-ray radiation Part II Positron Emission Tracking (PeTrack): the prototype and its evaluations Tong Xu, Marc Chamberland, Benjamin Spencer, Simon Massad Carleton University, Ottawa, Canada Outline Introduction Concept of PeTrack Simulation study and results The prototype the evaluation Conclusion External beam Radiation therapy http//www.stfranciscare.org Radiation delivery requirement Deliver tumour high radiation dose to Minimize radiation to healthy tissue around the tumour Accurate delivery of x-ray beam 3 Tricks... Trick #1: Focusing multiple beams Trick #2: Collimate the beam to the shape of tumour This method is called 3D-conformal radiation therapy (3D-CRT) 3D conformal Radiation therapy (3D-CRT) Shape the field following the outline of tumor Trick #3: Intensity modulation inside the field This is one step forward of 3D-CRT, with the addition of intensity modulation inside the field. Intensity modulated radiation therapy (IMRT) 3D-CRT Intensity is uniform inside the field Tumour Spinal cord IMRT Intensity is not uniform inside the field Accuracy of radiation therapy Significant development has been done in diagnose and delivery techniques (PET, SPECT, IMRT…) The tumor motion remains a limiting factor. The moving target Tumour moves due to: Respiration Cardiac beating Other visceral motions The tumor can move by more than 3 cm ! Three level of motion management None Radiation field Breath holding or Respiratory gating Real-time tumor tracking Motion management Breath holding Respiratory gating Breath holding Methods: – Self breath holding – Active breathing control device Limitations – Reproducibility (up to 6 mm residual motion) – Difficult for Lung cancer patient to tolerate Respiratory gating Breath normally! Uses: External markers Implanted internal markers Others – Spirometry – Temperature sensor – Strain gauge … Respiratory gating Gating Thresholds On Linac Beam Off Berbeco et al. 2005 External markers Tang et al. 2004 Correlation between external and internal motion Koch et al 2004 Unstable breathing Ozhasoglu et al. 2002 Phase shift Ozhasoglu et al. 2002 Internal markers Implanted in or close to tumor Invasive Provide exact location of tumor Internal marker tracked by x-ray Shirato et al. 2000 Image detector X-ray tube Radiation dose from the xray fluoroscopy Shirato et al. 2004 Up to 1.2 Gy skin dose per hour of treatment time Not feasible for intensity modulated radiation therapy – 20 – 30 minutes /fraction – large volume of normal tissue 25-30% of tumor dose! Calypso® 4D Localization System (EM marker) MRI artifacts of EM transponders X Zhu et al, 2009 RealEye tracking system Shchor et al 2010 RealEye tracking system Can only track one marker Can not be used for 10MV beam or higher due to induced radioactivity Tracking with Positron emission marker Miniature markers ( 0.8 mm) Labeled with positron emission isotopes (0.1 mCi) Track markers by detecting annihilation gamma PeTrack PeTrack is NOT PET Over-all image resolution of PET : 4 - 8 mm (Clinical Whole body PET) Can PET system locates an object with <1 mm accuracy? Yes, if the geometry of the source is known Find a point in 3D with the minimum summed distance to the coincident lines It is NOT image reconstruction ! Localize PeTrack marker Patient Detector Detector PeTrack system for tumor tracking Linac PeTrack detectors Positron emission Marker PeTrack Detector modules Detector module PeTrack marker and isotopes I-124 T1/2 (days) β+ Fraction As-74 Rb-84 4.2 18 32 23% 29% 23% Can be implanted with biopsy needle of size 18 Gauge (1.27 mm) The challenge The algorithm Classify the coincident lines using Mixture-of-Gaussians clustering technique Determine the position of each markers from its coincident line cluster Find the true location with iteration Initial estimation Computer simulation results Based on a Monte Carlo simulation package: GEANT4. Four markers were simulated Localization precision Localization error (mm) 1.5 1.0 0.5 0.0 0 50 100 150 200 Coincidient Lines per marker 250 Dynamic Thorax Phantom Phantom rod Marker Direction of motion RMSE (mm) AP 0.30 LR 0.20 IS 0.14 0.997 AP 0.42 0.636 LR 0.26 IS 0.17 0.997 AP 0.29 0.969 LR 0.25 IS 0.13 1 2 3 Average 0.24 3D RMSE (mm) R2 0.844 0.39 0.53 0.40 0.954 0.812 0.986 0.998 0.44 - The PeTrack Prototype BGO crystal and Position Sensitive PMT Single Marker Motion trace of the marker along the three coordinate axes 20 x y z Sine fit - x Sine fit - y Sine fit - z Position (mm) 10 0 -10 0 50 100 Time (s) 150 200 Adj. R2 Measured amplitude (mm) Expected value (mm) Error (mm) x 0.99 9.63 ± 0.05 10.00 -0.37 y 0.99 5.16 ± 0.04 5.34 -0.18 z 0.81 0.65 ± 0.02 0.67 -0.02 3D track of two markers 15 10 Z Axis 5 0 -5 -10 10 -15 15 5 0 10 -5 5 -10 0 -15 -20 -5 xis XA Positions of one of the marker X Y Z position (mm) 15 10 5 0 -5 0 10 20 30 Time(s) 40 50 60 Two markers precision Standard deviation of the distance between the two marks during the motion tracking: 0.73 mm Estimated precision: 0.52mm Conclusion PeTrack can perform tracking of multiple fiducial markers with sub-mm precision It is a potential technique for achieve hyperfractionation treatment for moving tumors. Acknowledgement Dr. Richard Wassenaar Nathan Churchill, University of Toronto Supported by Natural Sciences and Engineering Research Council of Canada Thank you! Tracking of a single Line marker Life time dose (0.1 mCi marker) Isotope 124I 74As 84Rb Half life (days) 4.2 18 32 dose (Gy) @ 5 mm (volume: 0.5cc) 2.6 9.0 18.6 dose (Gy) @ 10mm (volume: 4.2cc) 0.7 2.46 4.96 dose (Gy) @ 15mm (volume: 14 cc) 0.32 1.09 2.24 As compared with x-ray fluoroscopy dose Higher maximum dose Very small volume effected (~ 10 cc vs 1000 cc Can be implanted inside the tumor Precision 5.0 mm PET spatial resolution provides 0.5 mm localization precision With only about 100 events! Precision PET spatial resolution Number of coincident lines PeTrack data Predicted motion (100 ms latency) Predicted motion (200 ms latency) IS position (mm) LR position (mm) AP position (mm) 0 -10 Motion trace of marker #3 and predicted motion trace 80 70 10 0 0 5 10 Time (s) 15 20 2250 100 ms latency Mean = 0.0 S.D. = 0.8 Count 1500 750 0 200 ms latency Mean = 0.0 S.D. = 0.9 Count 1800 1200 600 0 -4 -2 0 2 1D prediction error (mm) 4 Distribution of the 1D prediction error 500 100 ms latency Count 400 Mean = 1.3 S.D. = 0.6 95th percentile (100 ms) = 2.3 mm 300 200 100 0 400 200 ms latency Mean = 1.4 S.D. = 0.7 Distribution of the 3D prediction error Count 300 200 95th percentile (200 ms) = 2.7 mm 100 0 0 2 3D prediction error (mm) 4 Latency (s) 1D pred. error (mm) 3D pred. error (mm) 0.1 0.0 ± 0.8 1.3 ± 0.6 0.2 0.0 ± 0.9 1.4 ± 0.7 Life time dose Activity = 0.1 mCi Sensitivity within the Field of view Frame based stereotactic neurosurgery http://www.elekta.com/healthcare_international_stereotactic_neurosurgery.php Fiducial-less tracking Schweikard et. al. 2004 Synthetic a serial of CT at different time points by deforming two CT scans : Inhale and exhale Registration of real-time x-ray projections with digitally reconstructed images from Synthetic CT scans Registration computing time: 5 -10 sec Accuracy depends on the deforming model of lung Physical Requirement of tumor tracked radiation therapy Track the tumor in real-time Predict the tumor position to account for the lag of delivery system Fast reaction of delivery system Current internal tracking techniques X-ray marker Sampling rate Precision Marker size 30 sec-1 EM marker 10 sec-1 0.5 mm 0.2 mm Φ0.8~1.6mm Φ1.8mm x 8. mm cylinder Radiation dose Upto 1.2 G/h Zero Correlation between external and internal motion Ozhasoglu et al. 2002 Complex tumor trajectory Ozhasoglu et al. 2002 Correlation coefficient (R) Koch et al. 2004 Spirometry Hoisak et al. 2004 Higher correlation (R= 0.51 - 0.99) than that of skin marker (R= 0.39 – 0.98) Difficult to tolerate Radiation dose from the xray fluoroscopy Shirato et al. 2004 External markers -1 Passive or active infrared skin markers Marker position tracked by camera in real time Linac gated by the position of external markers Linear accelerator generate pulsed x-ray Pulse frequency – 100 – 400 Hz Pulse width – 1 – 10 μs Blanking of PeTrack detector Expected data acquisition duty cycle > 80% PMT HV gating Expectation-Maximization -1 Expectation step. Compute the probabilities for all trajectories, n=1,…N, belonging to each cluster, k=1,…K p (i ) n,k (i ) (i ) (i ) ak G d (Tn , mk ) , k K (i ) (i ) (i ) a j G d (Tn , m j ) , j j 1 Expectation-Maximization -2 Maximization step. Update parameters N (i1) (i ) mk mk Vk a ( i 1) k (i ) 2 (i ) pn,k d (Tn , mk ) N k(i1) p n 1 N p n 1 (i ) n ,k n 1 N (i ) n ,k Previous works Gundogdu, 2005 Intended for industrial application Two particle was tracked Resolution 20 -30 mm The challenge Simultaneously tracking of three or more markers Distance between markers: a couple centimeters The existing algorithm for single particle tracking dose not apply Scatter rejection Patient R=2σ Expectation-Maximization iterations 1. Initial estimation 2. Expectation 3. Maximization 4. Clustering by the probability of each trajectory Update the position of markers Repeat step 2 and 3 until converge. Speed of the algorithm Four markers 400 coincident events 2.8GHz P4 20 ms/run Tumor position can be updated at a rate > 10 Hz Lift time dose for different treatment duration Required activity at the time of implanting Breath holding Methods: – Self breath holding – Active breathing control device Limitations – Reproducibility (up to 6 mm residual motion) – Difficult for Lung cancer patient to tolerate Respiratory gating Breath normally! Uses: External markers Implanted internal markers Others – Spirometry – Temperature sensor – Strain gauge … Respiratory gating Gating Thresholds On Linac Beam Off Berbeco et al. 2005 External markers Tang et al. 2004 Correlation between external and internal motion Koch et al 2004 Unstable breathing Ozhasoglu et al. 2002 Phase shift Ozhasoglu et al. 2002 Identify failed markers A failed marker should be identified automatically from the output of the algorithm ak Relative activity of marker # k k Root mean square distance form marker # k to its trajectories Identify failed markers k> 3 mm ak< 0.02 Identify failed markers with criteria The source of tumor motion Respiration Cardiac beating Other visceral motions Lung tumor motions trajectories Seppenwoolde et al. 2002 Internal marker tracked by x-ray Shirato et al. 2000 Image detector X-ray tube Internal marker tracked by x-ray Shirato et al. 2000 Positron emission and annihilation Positron Emission Tomography (PET) http://www.mni.mcgill.ca/cog/paus/techniques.htm PET image Physical limits on PET resolution Over-all resolution: 4 - 8 mm (Whole body PET) Humm et al, 2003 http://www.raytest.de/pet/clearPET/clearPET.html Three 22Na Markers Activity of 22Na: ~425 kBq/marker AP Rotation in AP/LR plane 25.6 mm 6.9 mm 11.3 mm LR Acrylic block Acrylic disk Active core (22Na) PET Image reconstruction http://depts.washington.edu/nucmed/IRL/pet_intro/intro_src/section4.html Yes! A single point source can be tracked with < 1 mm accuracy Park et al. 1993, Park et al. 2002 The algorithm Assuming the distance from a marker to its annihilation coincident lines follows a Gaussian distribution Standard deviation ~ system spatial k resolution Methods PeTrack simulation model Based on a Monte Carlo simulation package: GEANT4 Patient: Φ 30cm x 60 cm water phantom Distance from isocenter to detectors: 50 cm Detector: 40x40 array of 4x4x30 mm3 BGO crystals Energy resolution: 25% Spatial resolution ~ 4 mm PeTrack simulation model Marker: active 0.4 mm spherical core with a 0.2 mm thick gold shell Single marker simulation: – Sensitivity, scatter fraction, dose Four markers with I-124 were placed around isocenter: (0,0,0), (15,0,0), (0, 20,0), (0,0,20) (in mm) – Evaluate the algorithm Definition of a valid event (trajectory) Detected energies fall in the energy window (420600 keV) Coincidence has to be between detector A1 and A2, or between B1 and B2 Simulate the initial estimation error Error on the initial estimation – patient setup – respiration – marker migration Initial estimation is generated randomly around the true position – ± 5, ± 10 , ± 15 mm 1000 runs of the algorithm Definition of success marker and run Localized by the algorithm within 1.5 mm from its true position A successful run: – All four markers was allocated successfully Precision: – Mean error among 1000 runs from the true positions Run success rate 100 Run success rate (%) 90 80 70 60 50 ± 5 mm initial error ±10mm initial error ±15mm initial error 40 30 20 0 200 400 600 800 Coincident lineslines per per marker Total coincident run 1000 Marker success rate Mareker success rate (%) 100 95 90 85 80 ± 5 mm initial error ±10mm initial error ±15mm initial error 75 70 65 0 50 100 150 200 Coincident lines per marker 250 Number of runs with different number of Successful markers Initial error range (mm) ±5 ± 10 ± 15 All 4 markers are successful 997 3 markers are successful 2 markers are successful 1 marker is successful All 4 markers failed 985 777 3 15 144 0 0 75 0 0 4 0 0 0 Cardiac Beating Shirato et al. 2004 Yes! A single point source can be tracked with < 1 mm accuracy Park et al. 1993, Park et al. 2002, Sarah E. Palmer et al, 2006