SUBSURFACE IMAGING BMED 4800
Popular in Course
Popular in Biomedical Engineering
verified elite notetaker
verified elite notetaker
verified elite notetaker
Viva Pacocha II
verified elite notetaker
verified elite notetaker
This 163 page Class Notes was uploaded by Viva Pacocha II on Monday October 19, 2015. The Class Notes belongs to BMED 4800 at Rensselaer Polytechnic Institute taught by Badrinath Roysam in Fall. Since its upload, it has received 48 views. For similar materials see /class/224824/bmed-4800-rensselaer-polytechnic-institute in Biomedical Engineering at Rensselaer Polytechnic Institute.
Reviews for SUBSURFACE IMAGING
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 10/19/15
BM ED4800 ECSE48OO Introduction to Subsurface Imaging Systems Lecture 18 MRI II Kai Thomenius1 amp Badri Roysam2 1ChiefTechnologist Imaging Technologies General Electric Global Research Center ZProfessor Rensselaer Polytechnic Institute GE Global Research Center for SubSurface Imaging amp Sensing Page 1 Recap oLasttheiNec scussed Relation of MRI with respect to other imaging modalities MRI images oToday MR Imaging physics How do we make images with the spins Page 2 Some Properties of MRI Sca n ners 0Cost of siting an 0Cost of siting an MRI MRI scanner scanner cont Ranges from 100K Power Requirements to costs exceeding 480 V three phase those Of the o Gradient coils need 400 scanner amps 1000 V 0 Power hook Up RF pUlSGS 16 KW oAir conditioning 39SUpercondUCtmg BO Chilled water coils 320 Amps to get started 0 Room screening Page 4 Summary Magnetic Resonance Principles Some atomic nuclei have a property called quotspinquot OSpin gives the nuclei a magnetic moment oThese moments are randomly oriented When spins are placed in a magnetic field they align either with or against the field Nuclei Interact with Cl Magnetic Field With no magnetic field 3 Nuclei in random Nuclei align to the applied orientations field With a magnetic field Nuclear Orientations in BO 0 With 30 Spin up 2 Parallel Alignment M 0 Against B0 Spin down Anti parallel Alignment Z M E 39H39B HZB thZBo O Low energy Parallel orientation Bo High energy Antiparallel orientation Page 7 Magnet The Effect of Field Strength LT Ftinunn I g thumb Comporison of 15 and 3T performance Individudl Nuclei Precess about the Applied Field Lormor Equation Precession Frequencies f0 yZnB0 30T f0 12772 1Vle 15T f0 6386 1Vle Particle Spin Y2 It MHzT 1H 12 4258 23Na 32 1126 13C 12 1071 19F 12 4008 Page 11 In reality there are many nuclear spins in a sample V Page 12 In the rotating frame of reference the net magnetization can be represented by a single vector MZ The Rotating Frame Net Magnetization Z Z MZ Y Y X39 X Page 13 Energy absorption Impact of RF 39 I Rotating Frame Z of Reference M2 Y Y39 39V V X39 X39 MK 3 Before After Resonance olf you opply energy of the correct frequency to any system you get obsorption The opero singer and the gloss trick clf you opply energy of the correct frequency to nuclei they resonate and obsorb ie somejump from the lower ground stote to 0 higher excited stote Page 15 Energy emission 0 CICISSiCCII physics view Z Laboratory Frame of Reference Energy emission 0 CICISSiCCII physics view Z Laboratory Frame of Reference Why the Proton is used for Imaging The proton has the highest gyromognetic ratio of all the natural nuclei 0 Therefore the strongest signol Protons are present all over the body 0 Woter Fots lipids There are 0 lot of protons Eoch cubic mm of water hos about 2 gtltlO22 protons 0 Le 20000000000000000000000 Page 18 Signal amp RF Free Induction Decay A measurable magnetic resonance signal is produced as the transverse magnetism Mxy decays toward zero Signal decays to zero as the spins lose energy and lose phase coherence MKy precesses in the xy plane around the direction of B0 y Precession of Mxy vector induces current in RF C01 receiver coil Current in receiver coil decays to zero because of Loss of phase coherence Loss of energy Page 19 FID or Relaxation Different tissues relax at different rates 0 Liquid like tissues eg CSF relax slowly Soft materials like fat relax quicker Hard materials like bone relax too fast to be seen o In general pathological tissue relaxes slower that normal tissue 0 Relaxation is described in terms of two relaxation times 0 T1 is also called quotSpin lattice or Longitudinal relaxation T2 is also called quotSpin spin or Transverse relaxation oThese relaxation times are what is being imaged in MRI Page 20 T1 describes the return of longitudinal magnetization to equilibrium MA 0 Thne 39139 t T1 IVI0 39 IVI0 e 1 MA 0 Thne 439 t T1 Mzt M0 2MO e Page 21 T2 describes loss of phase coherence in transverse IVIxy magnetization WW t0 MXY MW 2 M0 exp t Tlme Page 22 Both processes occur simultoneously T1 90quot Hi P o g t17 t317 Page 23 T1 decoys for different tissues me 3 Signal intensity M best contrast 15 25 Pulse repetition time TR Page 24 How do we Select 0 Slice to be Imoged o Nuclei ore only excited by on RF pulse hoving the correct frequency 0 Magnetic field grodients moke the Lormor frequency depend on position 0 A limited bondwidth RF pulse opplied simultoneously with o mognetic field grodient will excite only those spins whose Lormor frequency is within the bondwidth of the RF pulse 0 Thus by opplying 0 selective RF pulse only the spins within 0 slice will be excited Spins whose Lormor frequencies ore out of the bond will not be excited Page 25 RF The Patient The magnetization vectors the sum of the individual lH nuclei will align along Bo like small compasses B0 Same Precession Frequency Homogeneous Field Page 26 RF The Patient All the magnetization vectors will interact with the RF field and rotate as long as the RF field is applied In this case by 90 degrees WM llllllIll RF Transmitted at the Larmor Frequency Cannot define Where the signal originates No Spatial Information Page 27 A Field Grodient Mokes the Lormor Frequency Depend upon Position 1500 T 63 861000 Hz Gradient in Z BZ 30 GZgt1ltZ Page 28 A Field Grodient Mokes the Lormor Frequency Depend upon Position 638 72000 Hz Xi 63 861000 Hz Gradient in X BX 2 Bo GXgt1ltX Page 29 RF amp Gradients Slice Selection at Isocenter While a slice selection gradient is turned on a shaped RF pulse is transmitted at 00 Only those protons precessing in a narrow range around 00 are excited rotated by the pulse RF is transmitted to the entire body but only the resonant wiare rotated Do 9 9 9 9 9 9 Gradient amplitude Isocenter j Page 30 RF amp Gradients MultiSlice Selection Multi slice Example RF pulses are transmitted at 3 different times and frequencies 3 different slices are selected A narrow RF frequency range around a central frequency calculated for each slice is transmitted to the entire patient Only those protons that resonate inthefrequencyrangethatis in gt 93 theslicelarerotated y quot quot 39 9 9 V quot quotlgtgt 4 i gt gt r 9939 gt gt 39 gt39gtgtgtgtgtgt39gt 9 V N quot 5 IQ 4 LA V gt t V gt gt r 3977 5 3 9 l 4 gt gt7 Q gt gt 9 Jp7 0 Do y22Gz Gradient amplitudes T T 00 39YZle j Isocenter Page 31 Slice Selection in MRI AB Z Excitation o A narrow frequency a A broader frequency band band or a strong or a weaker gradient gradient defines a thin defines a thicker slice slice Page 32 Now How do we scan the slice to get a 20 image Answer Use Gradients again Page 33 Visualization of the Gradient fields YGy Important B eld of all 39 sz gradients are in i Ldirectian 139 Page 34 Gradient gt Spatial rate of change i 0 Bx 33y aBZkGX Magnetic Field 3x 3x B VB Bx 53y quot0quotBZK Gy 3y 3y r max may 332 G I K Z 6 2 6 2 6 2 439 Gradient magnetic fields change the main magnetic field in a controlled and predictable pattern so the field is no longer homogeneous Page 35 Frequency Encoding During the echo signal acquisition the frequency encoding gradient is turned on It causes a known spatially dependent variation of precessional frequencies SO mm ween4mm dx 1971 foxirwr A 999 999 399 Y 399 99 999 1 X Gx is OFF Gx is ON Page 36 Phase Encoding Phase encoding occurs after siice seiectionexcitation The phase in the direction of the phase axis varies in space according to the area rnornenti of the phase encoding gradient 1 Iciy G 1517 r 2 k 2 51J39mxye J yydy 2 iny y y 899 7 1 7 leGy a kyyw y CD r 3 Gy is ON Gy IS OFF 39 r quot139 Page 37 A Simple Pulse Sequence Saturation Recovery SliCeSelection I39 RF See Reference 39 below for G 39 additional info S on ph Phase Encoding r f qumcy Gd encoding Frequency Encoding 39 L Readout G Signa4AM Source Page 38 Additional Info on Grodients Pulse sequence for MRI not to scale TRgtgtTEgtgtpulse lengths RF pulse 900 H I GZ slice selection H see Reference below for I GY phase en me I additional info change pulse length in NY steps 256 I on phase and frequency GX frequency encode eIlCOdlIlg read gradient I Signal FID acquisition time TE 4 pulse repetition time gt TR Source httpwwwloncsocukdeptsphysics tedchingpy336MRl short versiondocP 39 age Three Dimensional Imaging 0 Three dimensional imaging can be accomplished by adding another phase encoding dimension 0 During three dimensional data acquisition every collected data point carries information for the entire 3D image 0 3D imaging can provide isotropic voxels m m 1 Three slices from a three dimensional MRI data set Page 40 Summary We have discussed Proton spin oRelaxation Times oSpin dynamics Image formation in MRI oKey steps Image slice selection oGradients Page 42 Homework Lecture 18 0 Proposition MR Imaging works on o pulseecho mechanism with the RF coil as transmitter and receiver 0 Discuss the pros and cons of this proposition How is this similardifferent from the ultrasound pulse echo mechonism Page 43 Acknowledgments Thanks to Drs Charles Dumoulin and Thomas Foo of GE Global Research for the introductory slides o httpwwweradscommrimodhtm o httpradusuhsmilradhandoutsfletcherfletchersld025ht m o httpwwwcisriteduhtbooksmriinsidehtm o httpmriswmededuPhysWebOldmriphyslsldOOlhtm httpradusuhsmilradhandoutsfletcherfletcher httpdefiantsscuwocaJody webfMRl iDummiespdfs and potBasic MR Physics areypdf There are numerous additional sites on the web with excellent intros to MRI Page 44 Instructor Contact Information Badri Roysam Professor of Electrical Computer amp Systems Engineering Of ce JEC 7010 Rensselaer Polytechnic Institute 110 81h Street Troy New York 12180 Phone 518 276 8067 Fax 518 276 62612433 Email roysam ecser9iedu Website httQwwwr9ieduroysab NetMeeting ID for offcampus students 1281136180 Secretary Laraine Michaelides JEC 7012 518 276 8525 michalr9iedu quot Rensselaer why uni chug in lurl 39quot Page 45 Instructor Contact Information Kai EThomenius Chief Techno agist U trasaund amp Biamedica Of ce KW7C3OOA GE G aba Research Imaging Techno agies Niskayuna New Yark12309 Phone 518 38777233 Fax 518 38776170 Email thameniu crdgecam thamenius ecsergiedu Secretary Laraine Michaeh39des JEC 7012 518 276 78525 michargiedu RenSselaer GE Global Research m m mu m mum lge 46 BM ED4800ECSE4800 Introduction to Subsurface Imaging Systems Lecture 6 Introduction to CT Sconners Koi E Thomenius1 amp Bodri Roysom2 1Chief Technologist Imoging Technologies General Electric Global Research Center 2Professor Rensseloer Polytechnic Institute GE Glubal Research Center for Subsurface imaging amp Sensing Outline of Course Topics 0 THE BIG PICTURE PULSE ECHO METHODS What is subsurface sensing amp Examples imaging MRI Why a course on this topic EXAMPLE THROUGH A different sensing modality from TRANSMISSION SENSING the 9mm XRay Imaging BaSICS of MRI ComputerTomography MOLECULAR IMAGING 0 COMMON FUNDAMENTALS What iS it propagation of waves PET amp Radionuclide Imaging interaction of waves with targets o IMAGE PROCESSING amp CAD of interest Slidan 2 Modulation Transfer Function oanee aspects to xeray imaging Comth e xeray sources MTF 7 Cam th e xeray interactions with rnatter e detectors analog digitall Contrasttonoise ratio 0 Performance metrics MTF SNR CNR 7 I1 712711 i12 CNRDQEltargtlteray 7 7 7 N performance were given39 Detective quantum ef ciency eJusti cation for digital detectors Si was based on 0 ROC curves for h atnesis V 3 yp Un t VEM testing A 39 l 39 39 TP FP TN FN Naisepawer Exposure spec mR Incldmt energy 2 dms lgnlzimm gt LambertBeer Equation Review 2 0 86 where Hz is the xray intensity at the measuremen plane l is the xray intensity at the source plane Ia H 12 z is the distance between the e amp measurement planes pis the attenuation coef cient Z Also known as BeerrLarnbert or Beer s Law In general it IS a The goalofthe CT scanneris to apply the LamberteBeer function of 2 this Equation to nd the pivalues complicates matters across the cross section shim X Ray Imaging to CT Imaging 0 Standard Xray39s limitations 7 3D structures are coiiapsed into 2D images 7 Low soitetissue contrast great for ones 7 Not very quantitative 0 Xray CT 7 Take a iarge number of gtlt7rays at muitipie angies r Caicuiate the 3D image Simiiar hardware to ordinary gtltrray image oia siice 7 extendabie to 3D But heavy computationai ioad 0 1972 7 CT scanning invented by Godfrey i iounsiieid a UK scientist 7 Announced computed agtltiai transverse N scann ng 7 r initiaiiy known as CATscannerSthis caused predictabie cartoons to s ow u 393 7 Presented initiai crossrsectionai images oithe quot netc r The invention was shown to have exceiient diagnostic potentiai immediateiy 0 i iounsiieid shared the 1979 Nobei Prize with derived aiternative CT reconstruction aigorithms Uniortunateiy Dr i iounsiieid s company EMi iaiied to capitaiize on the invention and has since then ieit medicai imaging 0 Dr iiounsiieid passed away in 2004 hug mbelgnze mgmednumlaurutesl979hwuns dilecmre gar 51mm 6 Hounsfield s Apparatus Laboratory CT device Brain image CT Scanners o This is whatthey iaak iike aday o Basicaiiy rotating xiray tubes amp detectors 7 With a iot of computing power 0 CT scanners are being used or 7 AH types of medicai diagnostics 7 Airport iuggage inspection 7 Nondestructive evaiuation of materiais CT System Components Courtesy of Tom Toth GEHC 5mm m Note this kind of display comes a er a lot of graphics processing Image Reconstruction Problem CTimages generated by a reconstruction from projections I I quz Projections can be understood on the Z 0 8 basis 0 ambertBeer How can we generate an image 39 given a set of such projections The image shows a single measurement of attenuation through a brain section ln CT scanning many parallel surements are madeto form a projection ofthe attenuation Let us de ne our measurement a as I 1U giant CT Reconstruction Problem 0 Let us make 0 model of the unknown object n boxes of same size but different attenuation coefficients 0 Now opplying Lambert Beer we get iAxiAxiAx iAx 106 1 e z e s e n i iA C 2106 H gtlellt Slidem 13 CT Reconstruction Problem 0 Using our definition 9 we hove I n ln 0 Ax g I HM And in the limit as Ax goes to zero g T XVX 700 gtlellt 51mm 14 CT Reconstruction Problem 0 Our intuition tells us that a solution is possible 0 Here39s a quick demo to show that this indeed is the case Let us do four experiments in which we transmit four xray beams of intensity IE and receive four times with intensities I1 I2 I3 and I4 We can express the results mathematically with the LambertBeer relation I 101 Jul Mmr alvtix I I 1 gt 1 2 1 Jo In 1 M2IHZ 3WX gt 3 4 gt2 M3Il a lujx 13gtle 3 10 7 Thus We have 4 equations M4 7 Ink j 7 3 4 x 4 unknowns problem solved N a Slidem is CT Reconstruction Problem 0 Unfortunately the real world is not as kind as the example implies 0 For clinical utility we need at ledsta 512 by 512 pixel grid 0 Thus we would need at least 264144 equations to solve for that many unknowns 0 This is challenging to say the least when CT scanners were first introduced in the late 19705 this approach was use Images had to be calculated with fewer pixels people gave up on resolution so as to achieve reasonable reconstruction times 101 101 u I 1 2 1 12 3 4 gt 13L 14 Shdem m Projections amp Sinogroms P B t 0 WM Xrays Smogram Projection all rays in Sinogram 2D plot of direction 9 are all projections as a summed along the rays function of e and projection Width A er quot 39 39 quot TDh 1 indp htm Slidena Slidena 12 Basics 2D Fourier Transform 0 2D FFT of an image fgtlty 7 A good way to Uhderstahd CT recohstructtoh 7 ActuoHy Whote buhch of 1D FFTS 7 th the tmage shOWh most ehergy h tow frequehctes 7 Why 7 Assumthg mage stze of S what 5 the frequehcy thcremeh h FT 7 thh N ptxets What 5 the targest frequehcy h 7 Trahstormed tmage S W k75pace Spottotfrequehctes Fourer transform ofprqeeuon posslble w some nenng Theresult forms one 1ne othe ZD FFT othe ongmal 1mge Data aeqmsmon along one angle say them repeated for all desxred angles 51am 2n Fourier Slice Theorem fxygt Haj Fuv Fourier Slice Theorem can be used to derive a superior reconstruction approach Here s What it is 39 The 1D FT of a projection is equal to a radial slice of the 2D FT of the image With enough 1D F T s of the projections one can estimate the 2D FT of the image amp by taking the inverse the the nal image It also forms the basis for the ltered backprojection algorithm A er httpdolghinradiologyuioWaedugeSlidesC39I Ph 1indexhtm Shdenu 11 Recording scheme for CT scanning Note We actually record Intensity so curve should be inverse Simenu 22 Bockprojection with 0 Point Torget Imaging a point target Imaging a point target Bockprojection Alternative to Fourier Slice Recon Based on symmetry relations of 2D FT Once projection data are acquired one can begin the These steps are repeated for all projected angles Shdem 24 Introduction of Filtering w no ltering Impact of lter on Sinogram Backproj ection reconstruction Backproj ection reconstruction w lter compare images Shdem 25 Bockprojection Shepp amp Logon Phontom Shdem 25 Bockprojection Torso S u m mo ry 0 Introduction to CT sconners o Emphasis on reconstruction oigoritnrns 7 Basic probiern of reconstruction from projection 7 Simpie reconstruction using iineor equations 7 FourierSiiceTneorern 7 Bockprojection amp itered bockprojection 0 References 7 MtgLthuyerdurtmnuthEdu bgngueLENGGJB71122520RDC2520A rmiysis gdeE762 o 7 m dni hinmdmia uimuedu Esiid CW 5 indexhtm Homework Lecture 6 Algebraic Reconstruction Technique 0 Determine the attenuations in the four pixels as follows nor knowledge assume uniform attenuation afthefaurpixels taprawiu3an in art 7 Repeat tne process for tne vertical lineinte r s 7 Repeat until no runnerirnprovernents 4 6 7 Try ror oirrerent voioes preferably double digit 0 Compare result to one arrived at with the approach from Slide no 15 ART W35 the rst ISCOH algon39thm used may be making a comeback 51mm 2y Instructor Contact Information su Professor of E i JEC 70 nsseioer Polytechnic institute yr New Vork 12180 067 Eudri Roy m iectricoi Computer Ssystems Engineering 10 Re 110 a Street Tro Phone 51827678 Fax 518276762612433 Webs E u ra sab Secretary Loraine Michaelidesi JEC 7012 518 276 78523 michalrpi edu 51mm 2n Instructor Contact Information Kai E Thomenius Chwef Techno ogwst U trosound amp Bwomedwco Of ce KWiCEOOA GE G obo Research Phone 518 38777233 Fax 51813876170 Email thomemu crd ge com thomemus ecserg edu Secretary Loraine Michoehdes JEC 7012 518 276 78525 michorgi edu GE Global Research BMED4800ECSE48OO Introduction to Subsurface Sensing and Imaging Systems Lecture 17 MRI Kai Thomenius1 amp Badri Roysam2 1Chief Technologist Imaging Technologies General Electric Global Research Center 2Professor Rensselaer Polytechnic Institute GE Glubal Research Center for Subsurface imaging amp Sensing Summary 0 OCT has rown from a curiosity to a common y used tool in ophthalmic clinics worldwide 0 Many othernewer variations of OCT exist but we didn t cover them Frequency domain OCT 3D OCTetc Rapidly evolving field Trends point towards smaller handheld cheaper faster and more versatile OCT devices Sub micrometer resolution now possible Fundamentals of MRI Module 1 Quick Overview From the notes of 0 Charles Dumoulin PhD 0 Thomas Foo PhD and other sources March 21 2008 Magnetic Resonance Imaging Overview 0 What is Magnetic Resonance Imaging o What are the intrinsic parameters 0 Are there biohazard and safety issues 0 What are some of the clinical applications MRI amp Other Imaging Modalities Source X roy Detector o Sourcedetector geometry 0 Projective and computed tomography 0 Ultrasound o Sourcedetector at same location 0 Real time 2D imaging 0 Real time 3D4D imaging 0 Magnetic Resonance 0 Source of signal from within body 0 2D and 3D imaging Measured Parameters of Other Imaging Modalities 0 X ray 0 Electron density 0 Attenuation 0 Ultrasound 0 Variations in tissue compressibility amp density 0 Velocity of target tissue eg red blood cells 0 Positron Emission Tomography PE 0 concentration of radio labeled metabolites Measured Parameters of MRI 0 Nuclear spin density 0 Motion on molecular scale T1 T2 0 Motion on the microscopic scale diffusion perfusion o Macroscopic motion velocity acceleration etc 0 Chemical composition 0 Chemical exchange 0 Temperature 0 Mechanical and magnetic properties of tissue MRI has a very rich set of parameters for imaging Costs of Imaging Instrumentation X ray 0 Portable units inexpensive lt 100K 0 CT scanners 500K 12 million 0 Angiography suites 700K 17 million 0 Ultrasound 5K 150K Cost 0 Magnetic Resonance 500K 2 million Ultrasound Xeray CT MR Safety and Biohazards of Imaging Modalities X ray Ionizing radiation Morbidity associated with contrast agents 0 Ultrasound No safety or biohazard problems 0 Magnetic Resonance No biohazards However safety is an issue A Note on Magnetic Fields 0 Most often used unit Tesla Unit of magnetic flux density Equivalent to 1 Weber per sq meter or 10000 Gauss a 14 Tesla 10010 Gauss Magnet Safety The whopping strength of the magnet makes safety essential The magnetic field is neverturned off Things fly into the bore Even big things Source www howsmMomsmm Source hug NW simgiyghysies eum ing ob gds him Perhaps more dangerous are small items screw drivers scissors etc Magnet Safety hug www simplwhysics com ying ob ects hLml One can imagine the reaction of the buffer operator Removol ofo Chair from on MRI Patient Safety Anyone going near the magnet subjects staff and visitors must be thoroughly screened Subjects must have no metal in their bodies pacemaker aneurysm clips metal implants eg cochlear implants intrauterine devices IUDs some dental work fillings okay This subject was wearing a hair band With a 2 rhrh copperciarhp Left withhairbarid Right Without Subjects must remove metal from their bodies Source Jurge imich jewelry watch piercings coins etc wallet any metal that may distort the field eg underwire bra Subjects must be given ear plugs acoustic noise can reach 120 dB Clinical Applications of Magnetic Resonance Imaging Brain oVascuIar Stroke Neuro Cancer Periphera MS oJoints Parkinson s K Alzheimer39s etc 39 nees Shouder Spine Spinal cord Disks oAbdomen Emerging Applications of Magnetic Resonance Imaging 0 Cardiac Myocardial function Cardiac dynamics coronary artery ease o Interventional Image guided Biopsy Minimally invasive surger Vascular interventions 0 Functional MRI Why MRI in medical imaging 0 Water and fat are two major constituents of our bodies 0 Both have many hydrogen atoms 0 Hydrogen nuclei have an NMR signal 0 Hence 7 MRl primarily images the NMR signal from hydrogen nuclei 7 This assumes that the hydrogen density varies With tissue types amp clinical conditions The Origin ofthe MR Signal The Nucleus ofa Hydrogen Atom is 0 Charged Particle The Nucleus ofa Hydrogen Atom has a Nuclear Spin A Spinning Charged Particle eg the Hydrogen Atom will Produce 0 Magnetic Dipole E W11 Dipole magnetic moment Where D yisthe gyromogneticrotio 0 hour is Planck s constant 0 lroor is a unit vector MRI or NMR 0 MRI is a tomographic imaging technique used in medical imaging Magnetic resonance imaging 0 NMR is a spectroscopic technique used to obtain microscopic chemical and physical info about molecules Nuclear magnetic resonance 0 MRI produces images from thin slices of NMR signals 0 MRI was initially called NMR but this term was dropped due to the negative connotation associated with the term quotnuclearquot In 2003 there were appr 10000 MRI units in use amp 75M MRI scans were made httgWWWcis Iiteduhtbooksmrlinsidehtm developed as analytical tool New Technologies for Magnetic Resonance Imaging New magnet designs Higher fields Open geometry Faster and more powerful gradient subsystems MR compatible devices What happens in an MRI Scan 1 Put subject in big magnetic eld like 3 Tesla leave himher there 2 Transmit radio waves into subject about 3 ms 3 Turn off radio wave transmitter 4 Receive radio waves retransmitted by subject Manipulate re transmission with magnetic elds during this readoutinterval 10100 ms MRI is not a snapshot 5 Store measured radio wave data vs time Now go back to 2 to get some more data 6 Process raw data to reconstruct images 7 Allow subject to leave scanner this is optional me W Conventional Cylindrical Magnet quot H 07 Tesla Open Magnet L5 sunrem MR imaging MR imaging a m vmuxu m mum rm v be vqu r 5 mnmmm mm MR thermal imaging Contrast Agents in MRI Contrast Enhanced MR Angiogram Image courtesy of University of Melbourne MR imaging Future of Magnetic Resonance Imaging 0 MR will continue to grow at a rapid rate particularly outside the field of radiology This may be limited somewhat by our healthcare financial crisis 0 MR will displace many diagnostic methods in use today oThe cost of MR will continue to drop Summary 0We have discussed Relation of MRI with respect to other imaging modalities Safety of MRI related magnetic fields MRI images 0 Next time MR Imaging physics How do we make images with the spins Acknowledgments oThanks to Drs Charles Dumoulin and Thomas Foo of GE Global Research for most of these sides 0We have borrowed stuff from several sites on the web with excellent intros to MRI th39lwww hme limirh d 39 ndf httpwwweradscommrimodhtm th39lwwwri rit 39 quot 39 39M in htm ttglairtolnni lirla pdnRMF AlphRMF RIOQMnrld 39 BasicMRPhysicshtml Instructor Contact Information BOdI I Roysom Professor of Eiectricoi Computer amp Systems Engineering Office JEC 7010 Rensseioer Poiytechmc institute 110 8quot Street Troy New York 12180 Phone 518 27678067 Fax 5181276762612433 Email roysom ecsergi edu Website htt wwwr redoNro s b NetMeeting ID for offcampus students 128 113 61 80 Secretary TBD JEC 7012 Rousselaer Instructor Contact Information Kai E Thomenius Chief Technoiogist Uitrasound amp Biomedicai Office KWVCEOOA GE Gioboi Research imaging Technoiogies Niskoyuno New Vork 12309 Phone 51813877233 Fax 51813876170 Email thomeniu crd ge com thomemus ecsergi edu Secretary TBD Rensselaer GE Global Research n BMED 4800ECSE 48OO Introduction to Subsurface Imaging Systems Lecture 7 CT Scanning II Kai E Thomenius1 amp Badri Roysam2 1Chief Technologist Imaging Technologies General Electric Global Research Center 2Professor Rensselaer Polytechnic Institute GE Glubal Research Center for Subsurface Imaging amp Sensing Shdem 1 Outline of Course Topics 0 THE BIG PICTURE o PULSE ECHO METHODS Whatis subsurface sensing amp Exammes imaging MRI Why a course on this topic A different sensing modality from 0 EXAMPLE THROUGH the others TRANSMISSION SENSING XRuy Imaging Basics ofMRI Computer Tomography 0 MOLECULAR IMAGING 0 INTRO INTO OPTICALIMAGING Whatis It AND SENSING PET amp Radionuclide Imaging 0 COMMON FUNDAMENTALS o IMAGE PROCESSING amp CAD propagation of waves interaction ofwaves with targets ofinterest slide no 2 Recap of Last Lecture Introduction to CT scanners Emphasis on reconstruction algorithms Basic problem of reconstruction from projections Simple reconstruction using linear equations Fourier Slice Theorem Backprojection Si filtered backprojection Shdena z Quick XRay amp CT Review Measuring line integrals of attenuation coef cients u A av beam availing along Luis J is attenuated bv the object according to Beer s law for the photon intensities I IJ Ig ex 7 umdij tag The photon miensilles are prepzocessed m piojectlon data 1 I pl imi mom 10 m1 Thus we have a solve a Set of integral Equations for me attenuation cae ciems ii 39iquot deletion Source 39 39 quot Shdena A Quick X Roy amp CT Review Properties 0fthe attenuation coef cient u The object patient is described by the spatial distribution on the attenuation coefficient 1f The attenuation coefficient 171EIfp ZE is a function of 9 density material specific Z atomic number E energy of xray beam Usually the dependence of the beam energy E is only roughly considered by a beam hardening correction of the intensities 1 andE is set to an effective energyE gt EM Source share an s Houns eld Units CT numbers of tissue in Houns eld units HU 3000 6 T Lmnx39 I and mm I l 1 1 Mam umnbel A 1 V m I m1 9quot r Lung noon Source share an s Inverse Problems CT algorithms fall into a class of problems referred to as inverse problems Typical statement Given d generated that data in the CT settin data are the projections Jm find a 2D description of L t tissue attenuation that ra ed th t data A inverse Problems Often in posed 5m u mmewmmev mum mmtmm a No unique solution exists mmmmmmmm unstable Slidem 7 Getting into CT Details 0 Relation of e Radon Transform e FourierSliceTheorern e Filtered Backproiection o All otthese deal with makin animage ofylxyl from projections given by 1309 nony where P is the projection of rnxyl along L is the distance of Lfrom the origin 6 is the angle of Lwrt to the yaxis Slidem 2 Rodon Tro nsform o In 1917 Johonn Rodon published on exoct solution to this problem given below 0 None of the CT pioneers eg Hounsfield Cormock were owore of Rodon s work nor is the direct expression being used in ony commerciol sconners o This expression includes 0 derivotive of P016 given noisy x roy doto this would moke reconstructions chollenging o Other problems include the discontinuity ot t t this octuolly goes owoy if you do the 6Lintegrotion first xay 1 Ideidt39Lm 272 t t t where I xcos6ysin6 Slide no 9 2D Radon transform ER The analytical approach ofreconstruction hue integral by projections has to be done in the context of the Radon tmnsfonn 1 y normal x ector tinpgId21396f 57pxttF A fdlyp I5 Thus in the 2D case the Radon transfolm 93 is identical to the projection data p p015 HARM35 with projection angle 6 Source lm W39 quot1 394 yWWlTnmn39alt T html Slide no 10 w projection data p smogram the sinogram Source The representation of the don transform Ell10 in a oBdiagram is called A point in spatial domain appears as a sinusoidal curve in Smogram B Sinogrom Definition snag it w mm by de ning m m Ahmcmmnul Fnu rummn n nn 4 m Fourier Slice Theorem unruka quotm nhjerx an mpim xumvl prlyccnnnill o r IntIiquot 39nm fru mry nmmn mlcgml mvw imph u m w mum imcgml mu m mm mun 39 c m m Fauna sum Theorem in men rm szi comma he Fourier lmrnfnml of aim n r 7 n The Fquot mm m j39 jlm if m m My IIIc Ihux rmquot n m hinch deprndrm nu y w 3quot mm m 339 fix dy pmu a mi 5 m MunI nlung mm Innifnrm whim nu n 39 1quotva Sumquotmg ms in um uc nd P amrum ix I2 01 r 4 i p mmrum VI in mime1mm pm mu m hm uh luliuwing minimum Mn mesmm mp Henbweb em purdue edumalnlmpctCTLC1103 pas U snaem 12 Relationship between spatial Fourier and Radon domain M spa aldbmaip 2D Fourier n39ansfonn adon u39ansform radml 1D Fourier mxmawpm 9W 4 D Foll erddmain Radondomain Suurce mm waw xwrumrhexdelberg degmugsnggWWWITutanalCT html sham 13 Filtered Bockprojection coHect projecuon repeat for H The most Common reconstruchon o go thm re the Mtered backprojecuon metho Dvervwew FBP Reconstruchon Ste 5 1D Founer transform Proms m equency domam Mummy e CoHecL one projecuon drutew start processmg the am a projecuons backproject across rmage space Fuu er Transform rrverse 1D Fourrer transform 1 E auckprmectucruss rmug e Repeatfor 0H projecuon ang es amp Slaney Pnnclples Imagmg IEEE Press Ch 3 W a too html sham m Excellent reference Kak of Computerized Tomogaphlc h www slan or Filtered Bockprojection Last time we discussed the Fourier Slice Theorem 1D FT of each projection gives one radial line ofthe final img es 2D FT 9 The relation to FBP can be m y PM H understood from L I Each projection is nearly independent of each other 0 Can be more easily visualized in the frequency space 0 Only common point is at DC Thus we can add their quot contributions to each other 0OAH4 D lt Ivequenty domain Filter for FBP Before we can add the individual contributions ofthe Fourier transform we have to account for the non uniform sampling The ideal filter for this is the pie shaped wedges shown in Figure la Unfiltered processing would simply use response bl To emulate response a people have used the filter response c l e lf we assume there are K projections over 180 degrees then the width of the wedge will be 274 ml K eThe ramp filter shown in clis g g 397 Ln intended to model this response V 7 l K39 T ruquznzy domain Filtered Bockprojection Final reconstruction step is addition of the 2D inverse FT of each weighted projection This step is called backprojection smearing of each filtered projection over the image plane Advantages of this method over the Radon or 2D FFT methods Process can be started after first projection data is in Any interpolation is done as part of the backprojection Slide no 17 Filtered Bockprojection Finally the filtering process can be adjusted to bring about any modifications in the data The attached figure shows several such filters that have been used for this purpose Two examples are shown One can reduce higher frequency components by introducing a lower cutoff see Hann or Parzen filters Frequency Spahal cm 7 on lvequency l v quot FwKIwI Amplitude Frequency SheppALugclll quuenby HammlAg e Armin realism Slideno 18 Getting Perfect CT Data Most of the code in today39s scanners does not deal with reconstruction rather data optimization There is a real need to improve the quality of the raw data 7 Corrections for detector sensitivity variation Remember 9 yx lnioil Our measurements are critically dependent on uniform responses from all the detectors 7 Correction for beam propagation effects and detector ailings uAs it turns out doubling the path length seldom doubles the attenuation 7 things are actually worse 39 lower a 39 causing the remaining beam to have higher average energy Scattered radiation Detector nonlinearities each detector has its own nonlinearities Slide m w Matlab amp CT Reconstruction Matlab offers a forward radon and inverse iradon mefiles in our CT context R radonltheta ol contains an intensity image oTheta is a vector of projection angles 0R will give the resulting projections u when plotted as a 2D image it will be a sinogram 7 I iradonRtheta 39 39 0R is the projection data this is the point where we usually start oTheta is the same vector as above Slide m 20 Matlab amp CT Reconstruction 0 In addition to the radon m les Matlab gives you a SheppLogan Phantom m le P phantom512 imshowP 0 With those two statements we get this gray scale SheppLogan image 0 We can now take the radon transform of this theta 0179 quotA projection angles R xp radonP theta We can now display the resulting sinogram I l A 7 gure imagesctheta3xp R3 colormaphot colorbar xlabel39theta39 ylabel xprime39 Let s check how much worse things get with going back with the iradon m le P2 iradonR 2 Some Web Resources htt9Wwwonidorsted ufaridanaQregrintsfb M matlab code for filtered backprojection Designer Shepp Logan phantom Filter design possibilities Modified code available rom httQwwwecser9iedu censsisSSICourse Slide no 22 Some Web Resources W demos severol demos involving projections reconstructions etc W i Very nice demos for s iiiii a m ct imaging in general and l projections in particular Some Web Resources 0 ttpWwwctsimorg open source CT simulator 0 very nice tool to ploy games with various reconstruction para meters Hos to be dovvnlooded to run on your PC Completely menu driven opp CT Scanner Evolution 0 First Generation 1970 7 Parallel beam design 7 Onetwo detectors 7 Translationrotation 0 2nd Generation 1972 7 Small fan beam 7 Translationrotation 7 Larger noof detectors 0 3rd Generation 1976 7 Multiple detectors 7 Large fan beam 0 4th Generation 1978 e Detector ring 7 Source rotation 7 Large fan beam Slidenn 25 Fa n B e a m Data acquisition in fan beam geometry coilimated erav same Commercial scanner of the third generation acquire data in a fan beam geonietiy snag m 25 Reconstruction methods for fan beam geometry Rebinning Resainpling of fan beam to parallel beam 7 V fll ic 397 requires additional interpolations Application of reconsnuctiou for parallel beam requires waiting time for acquisition of many projections Refarmulation of the inverse Radon transfonn by a coordinate transfouuatiou 39mu parallel to fan beam geometry 7 Direct inversion slidenu 27 CT Sconner Evolution Multislice sconners Helicol GE or spirol Siemens sconners r Simultaneous Source rotation Table Translation Data Acquisition Electron Beom Tomography SinglerSIice Cr QuadrSIice CT One way tube and nine YEW at One way tube and multlple YDWS at detectors pimqu 1 cnannelat detectors pimqu A cnannela spatlal data suurau detectors atapatial data Manyihuusands at in a Single YDWaYE detectors in a 20 array slidenu 28 Multi slice CT 0 Going beyond cone beam command X I Area of keenest competition today I i paumi Md n ufd ctu rers a re I n coufdnwe ddding slices toddy continuous we are at 64 We 0 Next year who 9W me knows sud m 29 Multi slice CT Dates 1970 1978 Ci needie beam 395 pamai an beam I one siice delecior SlnglE detector 1993 future v mumJan beam i icons beam area deleclor 2003 03 05 odf Slidan 3n New Challenges w Helical CT 5EQIWQ quot4 ln Conventlonal CT 7 TD getunutherlmuge the ddmry lsmmEdm next lDCEIUDH Hellcal planar r Pduem Table mdves unywherefmm 1 7 1a mmsec CTcovers anonr geometry 7 D51 dlddnmms drama sdme uswlth cdnvermdndl Cl 7 An ddded lmerpdldtldn step Z39lnterpulutlun ls requlred Homework 70 I Vau wlll be supplled wltn a slnagram of an unknown lmage lunknawnnwllt7al shortervectors Atw tpolnt would you say the Image quality IS acceptable Mm 31 Homework 7b extra credit 0 Using the Forinodd filtered bockprojection code change filter parameters for 0 lower bondposs Demonstrote loss of spatial resolution Summary 0 Several detoils involving CT scanner operation were reviewed Distinctions among the major methods for image reconstruction In particular the role of the convolution filter in FBP was considered 0 CT related resources on the web were identified oThe vorious generations of CT sconners were defined Instructor Contact Information Badri Ro 5am Professor of EIectricaI Computer amp Systems Engineering Office JEC 7010 RensseIaer Poiytecnnic Institute 110 8 h Street Troy New York 12180 Phone 518 27678067 Fax 518 27662612433 Email roysam Qecsergiedu Website nttg www rgi eduNroysab NetMeeting ID for offcampus students 128 113 61 80 Secretary Loraine Michaeiides JEC 7012 518 276 78525 micnairgi edu Rensselaer M n m t quotm slide no 35 Instructor Contact Information Kai E Thomenius Chief Tecnnoiogist UItrasound amp Biomedicai Office KWVCEOOA GE Giobai Research Imaging Tecnnoiogies Niskayuna New Vork 12309 Phone 51813877233 Fax 51813876170 Email tnorneniucrd ge corn tnorneniusecseirgi edu Secretary Loraine Michaeiides JEC 7012 518 276 78525 micnairgi edu Rensselaer m nun 0 36 GE Global Research BMED4800ECSE4800 Introduction to Subsurface Sensing and Imaging Systems Lecture 9 Propagation of Waves II Kai Thomenius1 amp Badri Roysam2 1Chief Technologist Imaging Technologies General Electric Global Research Center 2Professor Rensselaer Polytechnic Institute GE Glubal Research Center for Subsurface ImagIng amp SensIng shde 1 Outline of Course Topics PULSE ECHO METHODS Examples THE BIG PICTURE MRI What IS EUbSUI face sensmg 8 A different sensing modality Imaglng from the others Why a course on this topic Basics ofMRl EXAMPLES THROUGH MOLECULAR IMAGING TRANSMISSION SENSING Whams it X39Ray Imaging PET amp Radionuclide Imaging ComputerTomography IMAGE PROCESSING amp Intro into Optical Imaging CAD 0 COMMON FUNDAMENTALS propagation of waves interaction of waves with targets of interest SIIde 2 Recap from last class oThe basic wave equation 2 i p c 1 02 0322 poK V21 0 Helmholtz equation 7 39a 52 Pt P0e 1 III a 2 k2P 0 Homogeneous Z 21 k2P With excitation Z 2 Slide 3 Recap Useful Relations oEquations on right are 2 27239 general k Applicable to most probes we discuss in this course a o For plane acoustic waves E C 1 lpoK f 2 a 0 Please note thatcis nota 27 function of frequency 0 As a consequence there is no 1 dispersion all frequencies move f at the same velocity Slide 4 Acoustic vs EM Waves Acoustic waves EM waves Wave types longitudinal or shear electromagnetic waves mechanical waves Transmission elastic medium no medium necessary requirements letherl Velocity of c 1 c 1 propagation paK 1mg Velocity of 1500 ms in water 30x108 ms in vacuum propagation Characteristic I0 impedance Z a K ml Amazingly similar for the rstorder wave equation 8 d 5 l e WaveMatter Interactions If a wave passes through a uniform medium we get the same wave unchanged at the other end Not very interesting in practice Medium is transparent to the wave Can t infer anything about the medium Non uniformities of the medium modify the wave as it passes This is interesting waves interact with the medium We can infer or even map the medium in terms of the wave matter interactions oSensing andor imaging is possible Slide 6 Acoustic Scattering There are two types of scattering we can call them quotcompressibilityquot and quotdensity based scattering Scattered field due to a point source 3 R 2m K K 3p 3pcos6 3rc K 297 p Duetmar mam me armquot a K ISLhe compressibility or some p ISLhe density ofthesource m Hdukemil 2 mm ednUlmsanndlkrs enadzZ himimcnonnnzznnnnnnnnnnnnnnnn Shde 7 WaveMatter Interactions Breast path Step n Time min us Shde 8 WaveMatter Interactions Chest path Step n Time mm us Slide 9 Simple WaveMatter Interactions Attenuation absorption Change in amplitudeenergy of the wave Reflection Waves bouncing off surfaces echoes Refraction Waves changing direction Diffraction Spreading of waves creating deviation from geometric paths Change in propagation speed Waves change phase Scattering Waves are redirected in many directions Dispersion Speed of propagation is a function of frequency Doppler Change in frequency caused by interaction with a moving object Slide 10 Contrast Generation 0 Each of these types of interactions is potentially a source of quotimaging contrastquot Changes in properties of the medium as we go from one point to another can be revealed by detectable differences in wave medium interactions 7 backgrwui backng I Michelson s Formula M max mm Weber s Contrast Formula 0 Contrast agents These are artificial substances that we can often inject They produce andor enhance contrast Slide 11 Attenuation Usually measured in units of quotdecibels ldBlquot 1 1Wattcm2 I2 10 Wattscm2 Attenuation 10xlogj Z 10x10g10 IOdB 1 Attenuation often changes by frequency of the wave The attenuation spectrum is characteristic for a given medium ie a spectral signature Attenuation frequency Slide 12 XRoy Absorption o Denser moteriols obsorb x rays more strongly 2 Zoe W Material Density p gcmz Air 00013 Water 10 Muscle 106 Fat 091 Bone 185 Source Richard Aston s online book Side 13 Xroy a bsorption Mass attenuation coefficient g cm2 g Xray Photon energy Water Air Bone Muscle keV 10 489 466 190 496 15 132 129 589 136 20 0523 0516 251 0544 30 0147 0147 0743 0154 40 00647 0640 0305 00677 50 00394 00384 0158 00409 60 00304 00292 00979 00312 80 00253 00236 00520 00255 100 00252 00231 00386 00252 150 00278 00251 00304 00276 200 00300 00268 00302 00297 300 00320 00288 00311 00317 Siide 14 Absorption based Contrast Agents Normal W 391 Difference X agent Injecte mm ood Vessels quot SW 15 Sample Numbers for Diagnostic Ultrasound Waves mghemequeney waves can xesolve smallex objects but penemsce 1ess mm lowerftequency waves shuns Attenuation is not all bad oOften we want to attenuate waves deliberately oSuch highly attenuating materials are called dampers We ll discuss them further when we talk about ultrasound imaging systems Slide 17 Reflections at Interfaces 0 Some oints on reflection p Zo Doc If Z2 21 there is obviously no re ection 0 Hence an impedance mismatch is Z Z needed to get a reflection If Z2 gt Zlthere is no polarity change however if Z2 lt Zlthe echo will be Zz Zl inverted 1 o This is necessary for there to be continuity in values at the boundary Upper video clip ZlZZ 05 Lower video clip ZlZZ 20 70 7 0 Fora nice discussion of this check39 httpphysicsusaskcahirose i 25animationre ectionanim 05 reflectionhtm Slide 18 Practical Application oln diagnostic ultrasound there is a large difference between the impedance of air and soft tissue A gel quotcoupling medium helps minimize reflection by bridging the impedance values of air and tissue quotImpedance matching Slide 19 Transmission at Interfaces 0 Some points on transmission If Z2 Zlthere ileO transmission T Z 1 0 Hence an impedance mismatch is needed to get a reflection ZZ Zl If Z2 gt Zlthere is no polarity changehoweverif Z2 ltZlthe echo T R 1 will be inverted o This is necessary for there to be continuity in values at the boundary Upper video clip ZlZZ 05 Lower video clip ZlZZ 20 Fora nice discussion of this check httpphysicsusaskcahiroseepZ 2Sanimationre ectionanim reflectionhtm 1 Refraction at Interfaces Refraction 39 39 I From Latin to turn aside a 39 quot At interfaces of media with differing propagation speeds Only occurs for oblique incidence Re ection amp Rgfracdon Changes the direction ofthe wave V1 lt V2 3 11 lt 12 V1 gt V2 311 gt 12 httplectureonlinecl msuedummgkgg13cd372htnt Slide 21 Diffraction Diffraction Sommerfeld39s 1894 definition any deviation of light rays from rectilinear paths which cannot be in interpreted as reflection or l 39 refraction i Closed form solutions are available for simple cases In practical complex situations we have to resort to solving the wave equation Diffraction httplectureonlinecl msueduNmmpkg213cd372htm Slide 22 Wave Superposition Interference onnsnumive Interference Destructive Interference u Propigalnn on 9 Direct 3 DPmpagaun in men Aw AAmplim e Aw c aquot 7 Wavelength 4 AW39AW ZAI AAmplltnde Av ivquot Resullam g Aw i c D i in MM Waves in Phase i C Paul Di emce D w Resmam CVlbrallon CVlh Avaiv mum er39mnh 4 Diwaliquot Wu AR c v C 1 o ltgt 1 80 i wavelength ltgt 360 2 ltgt Destructive Interference Imaging by Phase Differences interferometry Sensing wave Interference attern Reference wave Basic Idea The object affects the phase of the sensing wave relative to the reference wave If the two waves are brought together and allowed to interfere we can sense the phase change Very small effects fractions of awavelength can be sensed this Sid 24 i e Scattering Backscatter Waves redirected in many directions 39 0 Usually Scattering intensity is weaker ltV than re ect39on Surface Increases with frequency Speci c in terms of angular distribution Dependent upon the size and shape ofscatterers relative to A i o In ultrasound scattering J permits imaging of tissue boundaries that are not perpendicular to the incident Forward Scatter Slide 25 Dispersion Refractive index speed ofthe wave is a function of wave frequency Basis for prisms 39 L Key to spectral imaging unimum Jr39in mm m i39v VM L nu Doppler Example STATIONARY SOURCE transducer MOVING LISTENER red blood cell vcose mm Particle passes through 9 m approaching wavefronts V Stationary particle sees cT7t wavefronts in time T Moving particle sees CVCOSeT7 wavefronts in time T The frequency our particle sees is given by where Original transmit frequency Slide 27 EM and Acoustic Waves 0 Classical Acoustic and Electromagnetic wave phenomena are similar Reflection refraction diffraction interference polarization scattering some types etc 0 Differences arise when dealing with quantum phenomena Atomic phenomena are the basis for CT PET and MRI imaging Molecular phenomena are the basis for optical imaging Slide 28 EM and Acoustic Probes Electromagnetic S pectrum Gamma Rays 1 45 Acoustic Spectrum s 1quot Hays 1quot Infrasound lt20Hz 39 1 Ullravlolel 1039quot A E AUdlbIEI ZOHZ E m g 10quot l Vlslble m6 E c ZOkHZ Emquot i Infrared J g E Ultrasound gt20lltHz a 1039 quot1 E Mlcrowavcsl g 5g ZOMHZ r a 2 10 10quot 3 a Speed 3 5 swell 101 E Alli 330mS Radlo Waves 3 W 10 5 Water 1495 ms e 10 Figu39ez 109 Vlslhle tlghtsucenun Bone mS urrrshrur hm hm sum rum rm Figure 1 Slide 29 Physical Interactions of EM waves with Matter Wavelength Type ofinteraction Comment Range 10m 7 lrrreter cnange of nuclear Spln Nuclear Magnetic Resonance Nucleons absorbemit based on thelr Radxa magneth componentofEM Spin prepertv Frequency lrn e 1 em out Radxa Frequency magneth componentofEM prepertv wave more lrnportant lJm Change of onentatlonlrotatlon Mostly rotational effects Microwaves lelectnc component of EM Wave more lmportant uh r l 7 x t t r nonhuman t qlll t 1v llue lOnmrlooom Atthese u rtr pho c lxrray absorblng molecule by phamdxssacmnan or even produce pk ofl vldual atoms loooraugstrom photons wlll photororuze e ctrohsrr the outer shells where as lUUangstrom or shorter photons wlll photororuze hells amxamzatxan electrons m the lnner s 100prn and change of nuclear Mostly passes through smaller con guration Slide 30 Recap of the Lecture 0 Wave propagation at the heart of sensing and imaging systems Differential wave matter interactions are a primary source of imaging contrast 0We have noted some parallels between acoustic waves and electromagnetic waves Similar wave equations We ll discuss differences in greater depth later They are the basis for substance specific imaging Slide 31 Homework for Lecture 8 0 Using the data in slides 12 and 13 find the optimal x ray energies lvoltagesl that maximize the contrast between each pair of the materials bone muscle air muscle water airetcl Repeat the above exercise keeping in mind that higher energy x rays are more damaging Assume that damage to tissue is proportional to energy and determine the optimal energy levels for each of the above cases Slide 32 Instructor Contact Information Badri Ro 5am Professor of EIectricaI Computer amp Systems Engineering Office JEC 7010 Rensseiaer Poiytecnnic Institute 110 8 h Street Troy New York 12180 Phone 518 27678067 Fax 518 27662612433 Email roysam Qecsergiedu Website nttg www rgi eduNroysab NetMeeting ID for offcampus students 128 113 61 80 Secretary Laraine Michaeiides JEC 7012 518 276 78525 micnairgi edu Rensselaer n m t quotm Siide 33 Instructor Contact Information Kai E Thomenius Chief Technoiogist UItrasourid amp Biomedicai Office KW7C300A GE Giobai Research Imaging Technoiogies Niskayuna New Vork 12309 Phone 51813877233 Fax 51813876170 Email thameniu crd ge com tnomeniusecseirgi edu Secretary Laraine Michaeiides JEC 7012 518 276 78525 michairgi edu Rensselaer GE Global Research m 34 BM ED4800 ECSE48OO Introduction to Subsurface Imaging Systems Lecture 19 MRI II Kai Thomenius1 amp Badri Roysam2 1ChiefTechnologist Imaging Technologies General Electric Global Research Center ZProfessor Rensselaer Polytechnic Institute GE Global Research Center for SubSurface Imaging amp Sensing Recop Lost time we discussed Proton spins B0 and RF fields Flips of the mognetic moment Spin reloxotion T1 and T2 times System block diogrom oTodoy Spotio locolizotion of NMR responses New clinicol opplicotions A Typical MRI Sym 72 The Gradient Subsystem o Gradients provide spatial information They must 0 Be linear over the field of view Be accurately controlled 0 Gradients must move rapidly ie Rise to their required value quickly Settle at the value as soon as possible 0 Switching magnetic fields induce eddy currents in any conductor they penetrate Eddy currents oppose the field causing them Faraday39s law of induction 0 Therefore these currents must be minimized Compensation circuitry andor active shielding A Field Grodient Mokes the Lormor Frequency Depend upon Position 1500 T 63 861000 Hz Gradient in Z BZ 30 GZgt1ltZ Pulse bandwidth and slice thickness Frequency Hz A Gradient eg 1O Gcm 4260 Hzcm Narrow pulse I excites fat slice Position cm pJvisionpsvchumneducaomancoursesFa2006LecturesLecture2ppt Pulse bandwidth ond slice thickness Frequency Hz A Gradient e 10 Gcm 4260 Hzcm Strong gradient decreases slice thickness 4 gt Position cm pJvisionpsvchumneducaomancoursesFa2006LecturesLecture2ppt Pulse bondwidth ond slice thickness Frequency Hz A Center frequency I 4 determines slice position I l 1 Position cm Gradient e 10 Gc 4260 Hzcm A V th39l i inn n vrh umn edu r ODSl erfure ll Prhire nnf Typical Gradient Coil Designs Current Current Typical Gradient Coil Designs Current Typical Gradient Coil Designs Current i i i 39 Gradient Coil Designs Fingerprint Gradient Coils The Gradient Subsystem Two important parameters define the performance of a gradient subsystem Slew rate Maximum Gradient Strength Slew rate Determined by the voltage applied to the gradient coil Can be as high as 500 Volts Maximum Gradient Strength Determined by the current applied to the gradient coil Can be as high as 400 Amps Higher power subsystems give Faster Imaging leg Gradient echo Fast Spin Echo EPI Fewer Artifacts leg Better flow compensation l Diffusion weighted imaging Additional Info on Grodients Pulse sequence for MRI not to scale TRgtgtTEgtgtpulse lengths RF pulse 900 H I I l l GZ slice selection l l See Reference below for I GY phase en me I additional info change pulse length in NY steps 256 I on phase and frequency GX frequency encode eIlCOdlIlg read gradient I Signal FID acquisition time TE 4 pulse repetition time gt TR quot QQRIRIIDI k Hu r i n u doc Source AIAnAllanrlt Qtquot 39 39 I I39 39 w Slew Rate is an important parameter Gradient Amplitude or Gausscm mTeslametel O Time ms For highend systems gradient performance is limited by physiological stimulation 10 39 8 Gradient 6 39 GCm 4 P1IS39I mean 2 O A m 66PNSTmean 66 PNSTmean O I I I I I I I Peripheral 0 5 10 15 20 25 3O 35 Nerve Slew Rate St1mulat1on 39 Threshold Gcmms Radiofrequency RF Coils a RF coils act as o Transmitter ie apply the excitation Receiver ie detect the induced signal a Transmitter coils must 0 Provide a strong uniform field for a short period 0 Give a uniform excitation ie are usually volume coils o Receiver coils must Detect the weak NMR signal from the region of interest 0 Are therefore often surface or local coils 0 One coil can act as transmitter and receiver eg head amp body coils but frequently they are different 0 Multiple receiver coils can be combined to form a phased array MR Rodio Frequency RF Coils o Tronsmit coils con require 0 lot of power 0 Power requirements scole as the squore of the mognetic field strength 0 Lorger coils require more power At 15 Teslq 2 KW qmplifier used for the head coil 0 16 20 KW qmplifier used for the body coil 0 Receiver coils must be very sensitive o The MR signol induces uVolts in the coil 0 The best MR coils ore well tuned and matched to the lood o The MR sconner is ploced inside on RF screen room Keeps the RF excitotion pulse owoy from the surrounding environment Prevents imoge ortifocts due to interfering signals eg rodios computers etc What happens in on MRI Scan 1 Put subject in big magnetic field leave him there 2 Transmit radio waves into subject about 3 ms 3 Turn off radio wave transmitter 4 Receive radio waves retransmitted by subject Manipulate retransmission with magnetic fields during this readout interval 10100 ms MRI is not a snapsho 5 Store measured radio wave data vs time Now go back to 2 to get some more data 6 Process raw data to reconstruct images 7 Allow subject to leave scanner this is optional Source Robert Cox s web slides RF Coils amp Transmit Net magnetization teachers 2005MH rouqh qmde DDI RF Coils on Receive Free induction decoy FID v 4lA Details of the relaxation depend on the local environment We can exploit these differencs to emphasize different types of contrast in imags WWW A Simple Pulse Sequence Saturation Recovery SliCeSelection I39 RF Phase Encoding ed l Frequency Encodin N l L Readout G Signa4AM Signal Acquisition Raw Data The digitized data is stored in memory 256 or 512 data points equoi one view Eachlim 1 vim whichix 25 at 512 a point of am inalng m P hase Axis Frequency Axis KSpoce Representation in MRI The kspace is l The raw data collected after the scan 2 the spatialfrequency representation of the image 3 a grid of points each representing a complex number What determines the shape of the kspace The area under the gradient waveforms determines the shape of the kspace NOT the amplitude of the gradient pulse Kspace as a frequency domain representation ofthe image An axial brain slice Magnitude of K space data raw data 2 an inn an 2nn 25n an inn an 2nn 25n Center of k space reconstructed lowpass Q filtering 5n mu 15m 2mm 25m 5n mu 15m 2mm 25m Periphery of k space reconstructed highpass a filtering 5n mu 15m 2mm 25m 5n mu 15m 2mm 25m Functional Magnetic Resonance Imaging fMRI o The BOLD effect Some activities in the brain are well localized oActivation of brain tissue causes nearby capillaries to dilate o Capillaries carry a mixture of oxygenated and deoxygenated blood Oxygenated blood has a longer T2 than deoxygenated blood oThe dilation of the blood vessels exceeds the need for additional oxygen in the active tissue oThus brain activation causes a localized increase in 02 and a reduced T2 The increase in blood oxyhemoglobin is measure in f D Blood Oxygen Level e what we This is the BOL Dependent respon MRI Neuronal activity results in A in local An initial increase in oxygen consumption OlIi g to increased metabolic demand er a delay of 2 secs a large increase blood ow Whic overcompensates for the amount of oxygen being extracted Local increase in cerebral blood volume V613 l l mm m BOLD Blood Oxygen Level Dependent contrast Takes advantage of the different magnetic properties of oxyhemoglobin HbO and deoxyhemoglobin Hb Hb is paramagnetoand introduces an inhomogeneity into the nearby magnetic field whereas HbO is weakly diamagnefo and has little effect 90 o 6 o o o 0 0 5 W Agwi w 0 90 9 0 9 NormalFlow thnow oxyhemnanhin Demhmnvlnnn Neuronal activity gt local blood flow increases overcompensating for oxygen consumption gt oxygen level in venous blood is elevated gt larger MR signal LINEAR TRANSFORM MODEL What exactly does fMRI tell us Neural Activity llllllllll Measured MRI Signal 8 10 TlME sec WETEOEd HSHHANI OIWVNAGOWEH Buckner 2002 From sluggish Haemodynamic response to inferences on neural activity The central assumption the fMRI signal is approximately proportional to some measure of the local neural activity averaged over several millimeters and several seconds This is sometimes referred to as the linear transform model altered neuronal activity gtchanges in local hemodynamics gt fMRl signal How the three are related is unclear How to get on VIRI image response hox 39 ers comrnl magnet I In line dimensions stimulus control computer Radiolreqnency control computer ampli er fMRl image of the brain during finger tapping Image courtesy of University of Melbourne Role of fMRI studies brain 39 MRIfMRl studies brain 39 Another BOLD fMRI Experiment 0 I m not kidding about this one 0 U of Rochester researcher Dean Shibata did the following experiment With 14 patients 0 During MRI scanning the patients were told or readjokes of varying types Usually lamejokes intended to keep the patients from moving their heads 0 Why don39t sharks bite lawyers Professional courtesy Why did the golfer wear two pants Because he had a hole in one Increased brain activity was observed in specific sites in response to thesejokes Location of these sites varied on the nature and quality of thejokes These experiments were done on a GE Signa scanner Mognetic Resonance Spectroscopy MRS also knows os Chemicol Shift Imoging CSI Shift in PPM due to local chemicol environment is plotted vs signal to obtain 0 spectro Broin metobolites such as Choline Cho Creotine Cr and N AcetylAsportote NAA con be measured wwwuarizonaeduIeweagsggt Magnetic Resonance Spectroscopy oCholine Cho is o neurotronsmitter thot is increosed in tumors N ocetylosportote NAA is o neuronol metobolite thot is decreosed in tumors Liver Spectroscopy Increased Choline Metabolic map as cellular proliferation cell membrane synthesis and degradation Decreased Choline Metabolic atrophy a Effective therapy Pre TACE Post TACE Fundamentals of MRI Flow Imaging Magnetic Resonance Angiography MR flow measurement first demonstrated in 1960s Foundations of MRA laid in the 1980s Many methods developed since then All methods use changes in spin magnetization to discriminate vessels from surrounding tissue Longitudinal magnetization amplitude Transverse magnetization phase Cardiac Imaging MRA Challenges E EafFr ls l 16kHz Signal dropout due to turbulence Swallowing rt39f t Blood mov1ng faster than a 1 ac 667 Ginsec is never excited Velocity Sensitive Directions Sum Image 1 nun Phase sensitive methods detect only a single component of velocity at a time Phase contrast MRA can detect vessels that are smaller than the voxel size Simultaneous acquisition of soft tissue images and a 3D PC angiogram Acknowledgments Thanks to Drs Charles Dumoulin and Thomas Foo of GE Global Research for the majority of the slides httpwwweradscommrimodhtm httpradusuhsmilradhandoutsfletcherfletchers d025htm There are numerous sites on the web with excellent intros to MRI Lecture 19 Homework 0 Consider using on fMRI system as 0 lie detector Describe the steps you would undertake to design and implement such 0 device Conventionol polygrophs use physiologicol signols heort rote skin impedonce etc to recognize when d subject is lying How would you do this with M RI Instructor Contact Information Badri Roysam Professor of Electrical Computer amp Systems Engineering Of ce JEC 7010 Rensselaer Polytechnic Institute 110 81h Street Troy New York 12180 Phone 518 276 8067 Fax 518 276 62612433 Email roysam ecser9iedu Website httQwwwr9ieduroysab NetMeeting ID for offcampus students 1281136180 Secretary Laraine Michaelides JEC 7012 518 276 8525 michalr9iedu Rensselaer why uni cum in lurl 39quot Instructor Contact Information Kai EThomenius Chief Techno agist U trasaund amp Biamedica Of ce KW7C3OOA GE G aba Research Imaging Techno agies Niskayuna New Yark12309 Phone 518 38777233 Fax 518 38776170 Email thameniu crdgecam thamenius ecsergiedu Secretary Larar39ne Michaeh39des JEC 7012 518 276 78525 michargiedu Rensselaer GE Global Research m M mquot m mum
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'