Objectives 자기공명영상을이용한확산및관류영상의최신동향 장건호 경희대학교영상의학과 관류영상방법의최신동향을이해. 확산영상방법의최신동향을이해. 이두가지최근기술들이임상에적용할수있는가능성을이해. Content Perfusion( 관류 ) MRI Dynamic Contrast Enhanced (bolus passage) Dynamic Susceptibility Contrast (bolus tracking) Arterial spin labeling Diffusion( 확산 ) MRI Diffusion-weighted imaging Diffusion tensor imaging Fiber Tracking pmri: DCE or T1W PWI Common outputs: EES volume, CBF, CBV, Ktrans GD-DTPA 조영제 : 주입-> blood plasma 영역으로들어옴 정상세포 : BBB (blood brain barrier) 을통과하지못함 Tumor나 multiple sclerosis (MS) 등의질환에서는 EES(Extravascular Extracellular Space ) 로들어감. 조직의 T1 값을감소시켜 T1 강조영상에서신호강도를증가시키는역할 DCE: New Parameters constant rate: kep=ps/ve Transfer constant (Ktrans): BBB leakage 의경우 PS 의대용으로사용 flow 와 permeability 의감별을위함 Limited permeability model (ex, MS): Ktrans=PS (flow>>ps) Limited flow model (ex, tumor): Ktrans=flow (flow<<ps) Permeability Map 을구하기위한 free software: LUPE program Problems of the DCE Method Protocol: 3D Fast GE T1W sequence TR<7 msec, TE<1.5 msec, FA=30 Problems: 1. AIF 측정의정확성 2. 조영제가얼마나균일하게조직내에분포되는가 (tissue homogeneity) 3. 공간회상도및시간회상도를향상시키는문제와 (spatial and temporal resolution) 4. perfusion 및 permeability 를 angiogenesis 와관련시켜어떻게해석할것인가 5. Ktrans 및 ve 를임상에서손쉽게얻을수있는소프트웨어를만드는것. 1
pmri: DSE or T2*W PWI Common outputs: CBF, CBV, MTT, Peak height, Percent signal recovery 조영제의주입으로 Susceptibility( 자화감수성 ) 이증가하며, T2* 신호가감소하게됨. 이감소를시간에따른 T2* relaxivity contrast 변화 ( R2*) 로나타내게됨. 영상 : EPI sequence 를이용. Dynamic PWI Change in MRI signal due to passage of contrast agent Figure from Cerebral MR Perfusion. Sorensen and Reimer A. Pre-injection baseline B. Contrast arrival C. Peak signal change D. Recirculation E. Post-injection baseline A. Arrival time B. Maximum contrast concentration C. Full-width at half maximum Book by A.G. Sorensen & P. Reimer, 2000 Flow Quantification AIF: 정량적인 flow을얻기위해서는 brainfeeding artery에서조영제농도의변화를알아야됨 선형관계 : 조영제의농도와 R2* 의변화 실제로는비선형적 (quadratic relation) 관계임. Phase를이용한 flow quantification: 조영제의농도와 Phase 변화 : 선형적관계 flow quantification이개발되고있음. C m CBF measurement in DSC ( t) = CBF AIF( t) R( t) known unknown [Ostergaard et al. 1996] C m (t) : measured concentration curve in tissue CBF : Cerebral Blood Flow [ml/100g/min] AIF (t) : Arterial Input function R (t) : Residue function ( g / ml) 1 Cm( t) = ρ CBF( ml /100g / min) (min/ sec) AIF( t) R( t) k 60 H AIF And Partial Volume Effect (PVE) GE 영상에서는 voxel 크기가커서 PVE 의문제가발생 Artery 가 main magnetic field 에 parallel 평행인경우 : PVE 를어렵지만수정할수가있음. Artery 가 main magnetic field 와 non-parallel 경우 PVE 를보완하기가매우힘듬. local AIF 방법이개발되고있음 2
Deconvolution and Data Presentations Hematocrit 양의차이의보정 : 일정상수로처리 Dispersion 에대한 flow 값의오차가발생 Deconvolution 을위한최초발표된 singular value decomposition (SVD) 방법 최근개발된 Deconvolution 방법 : modified SVS Fourier method 및 maximum-likelihood expectation maximization (MLEM) Data presentations 방법 : ROI 각각의 voxel 에대하여 parameter 를구하고그 ROI 에대한평균하는방법 평균하고 parameter 를구하는방법이사용. Clinical Issues Percent signal recovery (PSR): Malignant glioma 와 metastasis 을구별하기위해이용 Metastasis 의경우 PSR 값이 capillary 에서의 GD- DTPA 의누수때문에감소 Glioma grading 을위해서는 rcbv 가일반적으로사용되어 rcbv 가클수록 high grade tumor. 예외 : Oligodendroglioma 의경우는 grade 에관계없이 vascularity 가증가해서 rcbv 를이용할수없음. Grading 을위한또하나의방법으로 Ktrans 을이용하고있다. pmri: ASL Common output: CBF ASL perfusion 영상 : RF 펄스를이용하여 arterial blood 의 longitudinal magnetization 를변화시켜얻는방법 PWI = Control - Labeled Development issues: blood labeling 효과를최대한도로높이기위함 MT effect 의최소화 : Static tissue 의기여도를줄임 ASL 방법 : pulsed ASL (PASL) continuous ASL (CASL) velocity selective ASL(VS-ASL) Vascular Territory Imaging (VTI) PASL va CASL PASL 방법 : 짧은시간동안 blood 를 labeling 예 : FAIR, PICORE, EPISTAR, UNFAIR, EST, TITL, DIPLOMA 및 IDOL CASL 방법 : flow driven adiabatic inversion 특별한하드웨어의지원이필요 PASL 방법에비하여어려움 SNR이높은것이장점 pseudo-casl이개발 3
RF G Pseudo-continuous Flow Driven Adiabatic Inversion 800 µsec label control Repeated vs. continuous RF More compatible with current MRI RF amplifiers on Allows product body coil excite/array receive 2-3X SNR improvement Improved efficiency (~80% versus ~70% for AM control) 2X SNR improvement Total 5X SNR gain With background suppression >10X SNR gain VSASL 방법 : blood 의최저속도를기준으로 (cutoff velocity Vc) 설정 뇌졸증환자와같은 cerebrovascular diseases 환자의경우 labeled blood 가영상영역으로흐르는시간이각각다르며 blood T1 값보다시간이길게될수있음 VS-ASL Wu WC, Wong EC, NeuroImage 2006 Garcia, de Bazelaire & Alsop, ISMRM 2005 VTI 영상법은 PASL 방법을이용하여일부 feeding artery 만을 labeling 하는방법임. Cerebralvascular disease 환자에서 stenoticvessel 들이여러개있을경우 carotid endarterectomy, stenting, 혹은 bypass 등의 intervention 시술을수행할경우에유용하게사용될수있음. VTI Green=left ICA Red=right ICA Blue=basilar artery Gunther M 2006, MRM 56 CBF and BOLD and CMRO2 ASL 영상법을이용하는경우 oxygen metabolism (CMRO2) 을얻을수있음 동시에 CBF와 BOLD 영상을얻고이를바탕으로 CMRO2을구할수있음 Dual-echo method: 1st echo=asl 2nd echo=bold CMRO2 Single-echo method: addition=bold subtraction=cbf CMRO2 Improved Reliability with DIPLOMA ASL Whole Brain ASL Perfusion MRI at 4T ICC DIPLOMA EPISTAR PICORE Global 0.81 0.78 0.78 Gray 0.80 0.75 0.76 White 0.80 0.79 0.79 Jahng et.al. Radiology 2005 Test-retest on 13 subjects ICC=Intra-class KSMRM correlation Spring 2007 coefficient Imaging Parameters: 2.4mm x 2.4mm; 3mm slices with 0.75mm gap; TR/TE=4200/11ms, 120 averages; 8 min 4
3D ASL with 1min Acquisition Intermission 3D Grase Seq 4.7x4.7x4m m with 26 slices 1min acquisition Diffusion Tensor MRI (DTI) Diffusion? Brownian motion of molecules (microscopic) Diffusion Tensor MRI (DTI) Diffusion? free diffusion restricted diffusion similar physical principle Einstein relation: r Alexander Leemans,Ph.D. 2 Diffusion (macroscopic) = 6 Dt water in water D = Diffusion Coefficient t = time r = displacement D is equal in all directions isotropic diffusion 3D 2D Alexander Leemans, Ph.D. D is not equal in all directions anisotropic diffusion Diffusion Tensor MRI (DTI) Diffusion tensor! mathematical framework to describe diffusion! ( e, λ ) 1 1 ( e, λ ) 3 3 ( e, λ ) 2 2 Alexander Leemans, Ph.D. D D D xx xy xz = Dyx Dyy Dyz D D zx Dzy D zz eigenvalue decomposition D= E Λ E e1 x e2 x e3 x λ 1 0 0 E = e 1y e2 y e3 y Λ= 0 λ 0 2 e 1z e2 z e3 z 0 0 λ 3 1 dmri: DWI Pulsed-gradient spin-echo (PGSE) echo-planar imaging (EPI) sequence b-value: 확산현상을결정짓는인자 trapezoidal gradient 을이용할경우 3 2 2 2 δ ε δε bij = γ GiG j δ + 3 30 6 5
Improvements Double refocusing Bipolar gradient Spin-Echo DTI Seq New PGSE-type sequence: Diffusion gradient 에의해발생되는 eddy current 에의한영상의변질을최소화 Double-refocused PGSE EPI sequence 가개발 Minimization of susceptibility artifact: Residual eddy-current 는 phase-encoding 방향으로 shearing, scaling, translation 등의영상변질을야기 phase array coil 의개발 MRM 2003 Two-compartment Model Diffusion Kurtosis Imaging, DKI DTI 영상이 Gaussian 분포를갖는다는가정함 실제인체에서는그러한분포만을형성하지는않음 Kurtosis: 통계분포에서사용되는 fourth moment 큰분포 (large displacement) 에매우민감한특성 DKI 영상 : 1 2 2 S = S0 exp bd + b D K 6 D=ADC, K=apparent kurtosis coefficient (AKC) Gaussian diffusion 에제한받지않음 K 값을얻기위해서는여러 b-value 에대한영상을얻은후에 fitting 을해야한다. 사람의 lung 과 normal human brain 에적용했음 ADC vs. AKC in Human Brain DTI 구조물질들의비등방성 (Anisotropy) 확산현상을기술 rotationally invariant scalar (RIS) 량 : 어떤방향으로영상을획득하든지관계없이확산값은변하지않음 환자간의확산을비교할수있는것이특징 RIS index: Isotropic diffusion: Trace (mean diffusivity) Anisotropic diffusion: FA, RA, 1-VR, LI diffusion-shape index (linear diffusion 과 spherical diffusion) Gray Matter White Matter Jensen JH, et al MRM 2005 Diffusion tensor 의 eigenvalue 와 eigenvector 를이용하여 white matter 의방향성을색으로나타낼수있다. Red=right-left, Green=anterior-posterior, Blue=head-feet 방향. 6
Diffusion Tensor Imaging (DTI) Fractional Anisotropy Mean Diffusivity E 1 E 2 T1-weighted E 3 Directionality Linear diffusion C li : λ1>> λ2 = λ3 Planar diffusion C pl : λ1 = λ2 >> λ3 Cylindrical diffusion C cy : λ1>> λ2 > λ3 Spherical diffusion C sp : λ1 = λ2 = λ3 Csp = 3 λ3/mean Anisotropic diffusion C an =C li +C pl =1-C sp Geometrical Expressions of Diffusion DTI directions DTI 영상을분석하는방법 ROI (region-of-interest) 방법 ROI 방법은쉽게사용할수있음 가정을증명하기위해서사용 분석하는사람의견해가들어갈요지가있음 Histogram 을이용하는방법 여러통계처리의반복에대한보상을하지않아도됨 선행되는가설이없는경우사용가능 D. Jones, 2006 ISMRM 표준공간의 group mapping 하는방법 voxel-based morphometry 분석 : SPM (Statistical Parameter Mapping) FSL(www.fmrib.ox.ac.uk)- TBSS(Track-based spatial statistics) 등과같은프로그램을이용하여하는것을 특정선행가설이없어도사용가능. Fiber Tracking (FT) or Tractography 이용 : functional network 의연구 neroanatomy 의연구 개발의중심점 : crossing-fiber 문제를해결하기위하여개발 본강의에서는각방법을구체적으로소개하는것을배제하고어떤방법들이있는가를살펴봄. Crossing fiber: Method 1 diffusion-tensor 를기초를둔방법 multiple tensor fitting 방법 FORECAST (fiber orientation estimated using continuous axially symmetric tensors) 현재임상에사용 대표적으로 Streamline: voxel 과 voxel 사이에 white matter 의연결을찾아내는데사용되고있다. DTIStudio, dtv Volume-1 이방법들은여전히 crossing 문제를해결하지못함 이를개선한 distributed connection 방법이개발 random walk diffusion model Monte Carlo method Front evolution 방법 Fast marching tractography (FMT) Probabilistic 방법. HARDI (High angular resolution DWI): 40-200 방향 b=2000 sec/mm2 이주로이용 7
Fiber Tractography (FT) virtual reconstruction of fiber pathways using DTI data Crossing fiber: Method 2 q-space acquisition scheme q=γδg/2π 로 reciprocal wave-vector 이다. 여러 q-value 을얻기위하여 b-value 에서와같이 diffusion gradient 의 strength 를변화하게됨 q-space 의특징은 DTI 에서가정한 Gaussian free diffusion 을사용하지않는다. Alexander Leemans, Ph.D. q-space 에기초를둔방법 : diffusion spectrum imaging (DSI) Q-ball imaging (QBI) persistent angular structure MRI (PAS-MRI) CHARMED (Composite hindered and restricted model of diffusion) DSI: q-space Wedeen, MRM 2005 Analysis Methods of FT 두번째로많은연구가진행되는것이얻어진 fiber track 을정량적으로어떻게분석할것인가? 단순히 fiber 를보여주는것외에 track 의 volume 을구하거나 fiber path 의 density 을구하는방법이소개되고있음. 또한 streamline 방법으로얻어진 fiber 의개수를얻음. 얻어진 fiber track 에서의 diffusion anisotropy 값을구하기. relaxation 시간을구하기 magnetization transfer ratio 값을구하는연구가진행되고있음 뇌의여러부위의 connectivity 에대한연구또한진행되고있음. Conclusions 최근들어뇌관류자기공명영상과뇌확산자기공명영상기술들이많이진보되었다. 뇌관류영상 : DCE 나 DSC-Perfusion MRI 경우는환자에주로사용 ASL-perfusion MRI 의경우임상적용이잘이루어지지않고있다. DWI 나 DTI 의영상 : 현재임상에일반적으로사용되고있으나 Fiber tractography: 뇌종양환자의수술전계획수립의경우에주로임상 임상연구나 neuroanatomy 및 functional connection 등의연구에주로이용 직접임상에사용하기전에많은선행연구가필요. 앞으로몇년내에는뇌확산및뇌관류영상이임상에보편적으로이용될것으로생각. Acknowledgements 대한자기공명의과학회 경희대학교부속동서신의학병원영상의학과 보건복지부보건의료기술진흥사업의지원 (A062284). 8