Chen Lin, PhD - Indiana University Bloomingtonmri/CE/slides/MR Perfusion... · Chen Lin, PhD DSC...

8
7/14/2009 1 MR Perfusion Imaging Techniques and Applications Chen Lin, PhD Indiana University School of Medicine & Clarian Health Partners Outlines Perfusion Modeling Methods of Perfusion measurements and clinical protocols. Perfusion Quantification and parametric maps. Examples of perfusion weighted imaging. Chen Lin, PhD Perfusion Deliver oxygen and nutrients to the cells Affected by pathological and physiological conditions, such as tumor angio-genesis, stroke and infarct, vascular wall changes. Chen Lin, PhD MR Perfusion Imaging Methods Dynamic Susceptibility Contrast (DSC) MRI Perfusion GRE-EPI (T2* weighted) SE-EPI (T2 weighted) Dynamic Contrast Enhanced (DCE) MRI Perfusion Spoiled Fast Gradient Echo (T1 Weighted) Arterial Spin Labeling Intra Voxel Incoherence Chen Lin, PhD DSC Perfusion Model and Parameters Blood Flow (ml of blood / gram of tissue / sec) Blood Volume (ml of blood / gram of tissue) Mean Transit Time (MTT) (sec) Chen Lin, PhD v Artery Vein CBF = CBV / MTT Capillary Bed Typical DSC Perfusion Protocol Single shot GRE-EPI or SE-EPI TE = 30 60 ms (GRE-EPI) or 50 80 ms (SE- EPI) TR = min. ( < 2 sec depends on number of slices ) TA = 90 - 120 sec. or ~ 100 time points Contrast dose = 0.1 0.2 mmol/kg Injection rate: 3 5 ml/sec with 20 ml saline flush. Chen Lin, PhD

Transcript of Chen Lin, PhD - Indiana University Bloomingtonmri/CE/slides/MR Perfusion... · Chen Lin, PhD DSC...

Page 1: Chen Lin, PhD - Indiana University Bloomingtonmri/CE/slides/MR Perfusion... · Chen Lin, PhD DSC Post-processing (Siemens) 1. Define a ROI for selecting AIF 2. Select an AIF among

7142009

1

MR Perfusion Imaging Techniques and Applications

Chen Lin PhD

Indiana University School of Medicine amp Clarian Health Partners

Outlines

bull Perfusion Modeling

bull Methods of Perfusion measurements and clinical protocols

bull Perfusion Quantification and parametric maps

bull Examples of perfusion weighted imaging

Chen Lin PhD

Perfusion

bull Deliver oxygen and nutrients to the cells

bull Affected by pathological and physiological conditions such as tumor angio-genesis stroke and infarct vascular wall changes

Chen Lin PhD

MR Perfusion Imaging Methods

bull Dynamic Susceptibility Contrast (DSC) MRI Perfusion

ndash GRE-EPI (T2 weighted)

ndash SE-EPI (T2 weighted)

bull Dynamic Contrast Enhanced (DCE) MRI Perfusion

ndash Spoiled Fast Gradient Echo (T1 Weighted)

bull Arterial Spin Labeling

bull Intra Voxel Incoherence

Chen Lin PhD

DSC Perfusion Model and Parameters

bull Blood Flow (ml of blood gram of tissue sec)

bull Blood Volume (ml of blood gram of tissue)

bull Mean Transit Time (MTT) (sec)

Chen Lin PhD

v

Artery Vein

CBF = CBV MTT

Capillary Bed

Typical DSC Perfusion Protocol

bull Single shot GRE-EPI or SE-EPI

bull TE = 30 ndash 60 ms (GRE-EPI) or 50 ndash 80 ms (SE-EPI)

bull TR = min ( lt 2 sec depends on number of slices )

bull TA = 90 - 120 sec or ~ 100 time points

bull Contrast dose = 01 ndash 02 mmolkg

bull Injection rate 3 ndash 5 mlsec with 20 ml saline flush

Chen Lin PhD

7142009

2

SE-EPI vs GRE-EPI for DSC PWI

bull SE-EPI signal (T2 dependent) is more specific to microvasculature while GRE-EPI signal (T2 dependent) is also sensitive to larger vessels

bull GRE-EPI provides greater sensitivity and coverage (more slices for same TR)

Chen Lin PhD

DSC Perfusion Imaging

Time

Signal

Percent Baseline at Peak

Time To Peak

Time To Minimum

Mean Transit Time

Negative Enhancement Integral

Baseline

Chen Lin PhD

DSC Post-processing (Siemens)

1 Define a ROI for selecting AIF 2 Select an AIF among the signal ndash

time curves for all the voxels in the ROI

3 Set time ranges in the AIF4 Calculate PWI maps

Chen Lin PhD

Perfusion Example

T2w

FLAIR

Chen Lin PhD

SWI

GRE-EPI

DSC Perfusion Maps

Relative Mean Transit Time

(relMTT)

Percent Baseline at Peak (PBP)

Time to Peak(TTP)

Negative Enhancement

Integral(NEI)

Chen Lin PhD

DWI PWI Mismatch in Stoke

DWI 4hr rCBV 4hr

Courtesy of Sorenson MGH Chen Lin PhD

7142009

3

Tumor versus Radiation Necrosis

T2w rCBV

Chen Lin PhD

Leakage and Permeability

Chen Lin PhD

Typical DCE Perfusion Protocol

bull Spoiled 2D or 3D gradient echo (SPGR or FLASH)

bull TE = min

bull TR = min ( typically 4 ndash 10 sec)

bull FA = 30 ndash 40 deg

bull TA = several minutes

bull Contrast dose = 01 ndash 02 mmolkg

bull Injection rate 3 ndash 5 mlsec with 20 ml saline flush

bull Measure baseline T1 with different FAsChen Lin PhD

DCE-MRI Evaluation

Qualitative - shape of signal intensity (SI) data curve

Semi-quantitative - indices that describe one or more parts of SI or [Gd] curves

Upslope gradient max amplitude washout rate or area under curve at a fixed time point

True quantitative - indices from contrast medium concentration changes using pharmacokinetic modelling

Chen Lin PhD

Type I(semi-necrotic with reactive changes)

Type II(viable tumor)

Type III(rapidly proliferating

tumor edge)

(Taylor and Reddick Adv Drug Del Rev 2000)

Qualitative Evaluation of DCE-MRI time Curve

Chen Lin PhD

DCE Perfusion Modelling

( )( ) ( )transt

p ep t

trans

ep

dC tK C t k C t

dt

Kk

Ve

Chen Lin PhD

Blood plasma

Cp

Ct = Ce Ve

Ve Ce

Ktrans

Kep

Tumor Extracellular

Space

NormalExtracellular

Space

7142009

4

Quantification of Contrast Concentration

R1(t) = R1(0) + C(t) Rbull R1(t) Longitudinal relaxation rate 1T1(t) at a given time

point t can be determined from the ratio of signal intensities between T1w and PDw images SI(T1w)SI(PDw)

bull R1(0) Longitudinal relaxation rate 1T1(0) at baseline ie before contrast arrival can also be determined from SI(T1w)SI(PDw) at baseline

bull R Relaxivity of the contrast agent ie Gd-DTPA

bull C(t) Contrast concentration at a given time t can be derived

bull Ct(t) Tissue contrast concentration curve

bull Cp(t) Blood contrast concentration curve arterial input function (AIF)

Chen Lin PhD

Quantitative analysis with pharmacokinetic modelling

bull Advantagesndash Whole curve shape is analysed

ndash Biologically relevant physiological parameters

ndash Independent of scanner strength manufacturer and imaging routines

ndash Enables valid comparisons of serial measurements and data exchange between different imaging centres

bull Disadvantagesndash Data acquisition and analysis is more complex

ndash Lack of commercial software for analysis

ndash Models may not fit the data observed

Chen Lin PhD

DCE Perfusion of Pancreatic Tumor

Chen Lin PhD

DCE Perfusion of Pancreatic Tumor

Chen Lin PhD

PWI with Arterial Spinning Labeling

bull Tagged blood flows through vasculature and exchanges with tissue

bull Change total tissue magnetization in slice (1-2) compared to control

bull ldquoControlrdquo ndash ldquoLabeledrdquo CBF

Artery Vein

Chen Lin PhD

Labeling Techniques

bull Continuous or Steady-state ASL (CASL)

ndash Saturation (Detre 1992)

ndash Inversion (Williams 1992) with flow-driven adiabatic inversion pulse

ndash Labeling with 2nd coil

bull Pulsed ASL (uses a bolus of labeled spins) (PASL)

ndash STAR TILT PICORE FAIR hellip

bull Velocity Selective ASL

Chen Lin PhD

7142009

5

CASL with Dedicated Labeling Coil

bull Reduce SAR

bull Reduce MT effect

bull Additional hardware

Imaging Plane

Labeling Coil

Chen Lin PhD

Typical PASL Design

M0 Label Control Label ControlLabelControl

Control - Label Control - Label Control - LabelDM

CBF = 3x108 l [mlg] a [] TI1 Exp(TI2T1blood) DMM0

Image

Acquisition

Module

TI2

TI1Labeling

Control

Module

Bolus Cut-off Module

Chen Lin PhD

Chen Lin PhD

QUIPSS amp QUIPSS II

bull QUantitative Imaging of Perfusion using a Single Subtraction

bull Use saturation pulses to shave the leading or trailing edge of the bolus so the transit time is the same

Image

Acquisition

Module

TI2

TI1Labeling

Control

Module

900 900

ASL Image Acquisition Mode

bull 2D single and multiple slices

bull 3D (long acquisition time)

Chen Lin PhD

ASL Image Acquisition Methods

bull GRESE EPI (STAR + EPI Edelman 1994 FAIR + EPI Kim 1995)

bull HASTE (STAR + HAlf-fourier Single-shot Turbo spin Echo Chen 1997)

bull GRASE (FAIR + GRadient echo And Spin Echo Crelier 1999)

bull TrueFISP (FAIR + True Fast Imaging with Steady-state Precession Boss 2007)

Chen Lin PhD

PASL Issues and Optimization

bull SNR and Susceptibility artifacts

ndash Use short TE amp ETL

bull T1 Relaxation effect amp Transit Time Difference

ndash Decay of magnetization from the tagging location to imaging location

ndash The transit time of leading and trailing edge of the labeled bolus are different

bull Arterial Signal

ndash Apply a small bipolar diffusion weighting gradient b ~ 2 smm2

Chen Lin PhD

7142009

6

PASL Issues and Optimization

bull Physiological Noise

ndash Reduction with background suppression

ndash Model and correct the physiological fluctuation with statistical analysis (GLM)

bull MT effect

ndash Suppression at imaging location from off-resonance effect of labeling pulse Asymmetric with respect to frequency shift

bull Eddy Current Compensation

ndash Maintain the similar gradient waveform between tagging and control

Chen Lin PhD

MT Effect amp Labeling Schemes

bull STAR (Signal Targeting with Alternating Radio Frequency)

bull PICORE (Proximal Inversion with a Control for Off-resonance Effect) (Wong 1997)

bull TILT (Transfer Insensitive Labeling Technique) (Gobay 1999)

bull FAIR (Flow-sensitive Alternating Inversion Recovery) (Kim 1995)

Chen Lin PhD

PICORE

Tagging

Control

1800

1800

wt = wi + Dw

wc = wi + Dw

Gt = 0

Chen Lin PhD

FAIR

Tagging

Control

1800

1800

wt = wi

wc = wi

Gt = 0Chen Lin PhD

ASL Applications

bull Quantification of rest state CBF and study the perfusion changes from a disease or treatmentintervention

ndash Tumor Stroke Alzheimerrsquos Disease vascular disorder etc hellip

ndash Infant brain development (Wang Neuroimage 2008) amp Aging (Biagi JMRI 2007)

ndash Uncoupling contributions to BOLD fMRI

Chen Lin PhD

AVM

T2wDSCrMTT

ASLrCBF

DSCNEI

Chen Lin PhD

7142009

7

T2-weighted

Ischemia (RgtL)

Pre-op ASL

Hypo-perfusion (RgtL)

Post-op ASL

Bilateral increase

ASL perfusion correlated better with cognitive deficits than structural MRI

Jefferson AJNR 2006

Intracranial Stenosis

Chen Lin PhD

Du A T et al Neurology 200667125-1220

Frontotemporal Dementia (FTD) and Alzheimer Disease (AD)

FTD and AD display different spatial distributions of hypo-perfusion Chen Lin PhD

ASL Perfusion based fMRI

bull Perfusion fMRI ie Quantification of functional related dynamic CBF changes

ndashConventional stimulustasks such as sensory-motor memory speech tasks similar to those in BOLD fMRI

ndashPharmacological stimulus such caffeine or cocaine

ndashBehaviorenvironment changes such as sleep deprivation pain

Chen Lin PhD

Xu G et al Neurology 2007

CBF changes of congnitively normal (CN) amp amnestic mild cognitive impairment (aMCI) groups

A significant regional CBF percent increase occurred in the region of the right parahippocampus during the sustained memory-encoding period in the CN group but not in the aMCI group

Chen Lin PhD

ASL versus BOLD for fMRIbull Much lower SNR

1-2 signal change in ASL versus 5-10 in BOLD

bull Direct measure of CBF changesBOLD is coupled with CBF CMRO2 CBV

bull More localized and quantitative

bull Better long-term stability and reproducibilityLow frequency fMRI

bull Insensitive to magnetic susceptibility effectsUse short TE and susceptibility insensitive image acquisition

bull Less inter-subject variabilityAllows group analysis long term and multisite study

Chen Lin PhD

ASL versus DSC

bull Lower SNR and therefore typical acquired with lower resolution

bull More susceptible to physiological variation

bull Less coverage

bull Non-invasive

Chen Lin PhD

7142009

8

ASL Outside of the Brain

bull Kidney (Warmuth JMRI 2007)

bull Muscle (Frouin MRM 2006)

bull Lung (Bolar MRM 2006)

Chen Lin PhD

Getting more from Diffusion Attenuation Curve

Ln(S)

b

Perfusion

S = Sb=0 [ (1-f) endashbD + f endashb(D+D) ]

b = 50

b = 500

Chen Lin PhD

Intra Voxel Incoherence Imaging

Luciani A et al Radiology 2008 Dec249(3)891-9

DWI b=50 D

D fChen Lin PhD

Thank you

Page 2: Chen Lin, PhD - Indiana University Bloomingtonmri/CE/slides/MR Perfusion... · Chen Lin, PhD DSC Post-processing (Siemens) 1. Define a ROI for selecting AIF 2. Select an AIF among

7142009

2

SE-EPI vs GRE-EPI for DSC PWI

bull SE-EPI signal (T2 dependent) is more specific to microvasculature while GRE-EPI signal (T2 dependent) is also sensitive to larger vessels

bull GRE-EPI provides greater sensitivity and coverage (more slices for same TR)

Chen Lin PhD

DSC Perfusion Imaging

Time

Signal

Percent Baseline at Peak

Time To Peak

Time To Minimum

Mean Transit Time

Negative Enhancement Integral

Baseline

Chen Lin PhD

DSC Post-processing (Siemens)

1 Define a ROI for selecting AIF 2 Select an AIF among the signal ndash

time curves for all the voxels in the ROI

3 Set time ranges in the AIF4 Calculate PWI maps

Chen Lin PhD

Perfusion Example

T2w

FLAIR

Chen Lin PhD

SWI

GRE-EPI

DSC Perfusion Maps

Relative Mean Transit Time

(relMTT)

Percent Baseline at Peak (PBP)

Time to Peak(TTP)

Negative Enhancement

Integral(NEI)

Chen Lin PhD

DWI PWI Mismatch in Stoke

DWI 4hr rCBV 4hr

Courtesy of Sorenson MGH Chen Lin PhD

7142009

3

Tumor versus Radiation Necrosis

T2w rCBV

Chen Lin PhD

Leakage and Permeability

Chen Lin PhD

Typical DCE Perfusion Protocol

bull Spoiled 2D or 3D gradient echo (SPGR or FLASH)

bull TE = min

bull TR = min ( typically 4 ndash 10 sec)

bull FA = 30 ndash 40 deg

bull TA = several minutes

bull Contrast dose = 01 ndash 02 mmolkg

bull Injection rate 3 ndash 5 mlsec with 20 ml saline flush

bull Measure baseline T1 with different FAsChen Lin PhD

DCE-MRI Evaluation

Qualitative - shape of signal intensity (SI) data curve

Semi-quantitative - indices that describe one or more parts of SI or [Gd] curves

Upslope gradient max amplitude washout rate or area under curve at a fixed time point

True quantitative - indices from contrast medium concentration changes using pharmacokinetic modelling

Chen Lin PhD

Type I(semi-necrotic with reactive changes)

Type II(viable tumor)

Type III(rapidly proliferating

tumor edge)

(Taylor and Reddick Adv Drug Del Rev 2000)

Qualitative Evaluation of DCE-MRI time Curve

Chen Lin PhD

DCE Perfusion Modelling

( )( ) ( )transt

p ep t

trans

ep

dC tK C t k C t

dt

Kk

Ve

Chen Lin PhD

Blood plasma

Cp

Ct = Ce Ve

Ve Ce

Ktrans

Kep

Tumor Extracellular

Space

NormalExtracellular

Space

7142009

4

Quantification of Contrast Concentration

R1(t) = R1(0) + C(t) Rbull R1(t) Longitudinal relaxation rate 1T1(t) at a given time

point t can be determined from the ratio of signal intensities between T1w and PDw images SI(T1w)SI(PDw)

bull R1(0) Longitudinal relaxation rate 1T1(0) at baseline ie before contrast arrival can also be determined from SI(T1w)SI(PDw) at baseline

bull R Relaxivity of the contrast agent ie Gd-DTPA

bull C(t) Contrast concentration at a given time t can be derived

bull Ct(t) Tissue contrast concentration curve

bull Cp(t) Blood contrast concentration curve arterial input function (AIF)

Chen Lin PhD

Quantitative analysis with pharmacokinetic modelling

bull Advantagesndash Whole curve shape is analysed

ndash Biologically relevant physiological parameters

ndash Independent of scanner strength manufacturer and imaging routines

ndash Enables valid comparisons of serial measurements and data exchange between different imaging centres

bull Disadvantagesndash Data acquisition and analysis is more complex

ndash Lack of commercial software for analysis

ndash Models may not fit the data observed

Chen Lin PhD

DCE Perfusion of Pancreatic Tumor

Chen Lin PhD

DCE Perfusion of Pancreatic Tumor

Chen Lin PhD

PWI with Arterial Spinning Labeling

bull Tagged blood flows through vasculature and exchanges with tissue

bull Change total tissue magnetization in slice (1-2) compared to control

bull ldquoControlrdquo ndash ldquoLabeledrdquo CBF

Artery Vein

Chen Lin PhD

Labeling Techniques

bull Continuous or Steady-state ASL (CASL)

ndash Saturation (Detre 1992)

ndash Inversion (Williams 1992) with flow-driven adiabatic inversion pulse

ndash Labeling with 2nd coil

bull Pulsed ASL (uses a bolus of labeled spins) (PASL)

ndash STAR TILT PICORE FAIR hellip

bull Velocity Selective ASL

Chen Lin PhD

7142009

5

CASL with Dedicated Labeling Coil

bull Reduce SAR

bull Reduce MT effect

bull Additional hardware

Imaging Plane

Labeling Coil

Chen Lin PhD

Typical PASL Design

M0 Label Control Label ControlLabelControl

Control - Label Control - Label Control - LabelDM

CBF = 3x108 l [mlg] a [] TI1 Exp(TI2T1blood) DMM0

Image

Acquisition

Module

TI2

TI1Labeling

Control

Module

Bolus Cut-off Module

Chen Lin PhD

Chen Lin PhD

QUIPSS amp QUIPSS II

bull QUantitative Imaging of Perfusion using a Single Subtraction

bull Use saturation pulses to shave the leading or trailing edge of the bolus so the transit time is the same

Image

Acquisition

Module

TI2

TI1Labeling

Control

Module

900 900

ASL Image Acquisition Mode

bull 2D single and multiple slices

bull 3D (long acquisition time)

Chen Lin PhD

ASL Image Acquisition Methods

bull GRESE EPI (STAR + EPI Edelman 1994 FAIR + EPI Kim 1995)

bull HASTE (STAR + HAlf-fourier Single-shot Turbo spin Echo Chen 1997)

bull GRASE (FAIR + GRadient echo And Spin Echo Crelier 1999)

bull TrueFISP (FAIR + True Fast Imaging with Steady-state Precession Boss 2007)

Chen Lin PhD

PASL Issues and Optimization

bull SNR and Susceptibility artifacts

ndash Use short TE amp ETL

bull T1 Relaxation effect amp Transit Time Difference

ndash Decay of magnetization from the tagging location to imaging location

ndash The transit time of leading and trailing edge of the labeled bolus are different

bull Arterial Signal

ndash Apply a small bipolar diffusion weighting gradient b ~ 2 smm2

Chen Lin PhD

7142009

6

PASL Issues and Optimization

bull Physiological Noise

ndash Reduction with background suppression

ndash Model and correct the physiological fluctuation with statistical analysis (GLM)

bull MT effect

ndash Suppression at imaging location from off-resonance effect of labeling pulse Asymmetric with respect to frequency shift

bull Eddy Current Compensation

ndash Maintain the similar gradient waveform between tagging and control

Chen Lin PhD

MT Effect amp Labeling Schemes

bull STAR (Signal Targeting with Alternating Radio Frequency)

bull PICORE (Proximal Inversion with a Control for Off-resonance Effect) (Wong 1997)

bull TILT (Transfer Insensitive Labeling Technique) (Gobay 1999)

bull FAIR (Flow-sensitive Alternating Inversion Recovery) (Kim 1995)

Chen Lin PhD

PICORE

Tagging

Control

1800

1800

wt = wi + Dw

wc = wi + Dw

Gt = 0

Chen Lin PhD

FAIR

Tagging

Control

1800

1800

wt = wi

wc = wi

Gt = 0Chen Lin PhD

ASL Applications

bull Quantification of rest state CBF and study the perfusion changes from a disease or treatmentintervention

ndash Tumor Stroke Alzheimerrsquos Disease vascular disorder etc hellip

ndash Infant brain development (Wang Neuroimage 2008) amp Aging (Biagi JMRI 2007)

ndash Uncoupling contributions to BOLD fMRI

Chen Lin PhD

AVM

T2wDSCrMTT

ASLrCBF

DSCNEI

Chen Lin PhD

7142009

7

T2-weighted

Ischemia (RgtL)

Pre-op ASL

Hypo-perfusion (RgtL)

Post-op ASL

Bilateral increase

ASL perfusion correlated better with cognitive deficits than structural MRI

Jefferson AJNR 2006

Intracranial Stenosis

Chen Lin PhD

Du A T et al Neurology 200667125-1220

Frontotemporal Dementia (FTD) and Alzheimer Disease (AD)

FTD and AD display different spatial distributions of hypo-perfusion Chen Lin PhD

ASL Perfusion based fMRI

bull Perfusion fMRI ie Quantification of functional related dynamic CBF changes

ndashConventional stimulustasks such as sensory-motor memory speech tasks similar to those in BOLD fMRI

ndashPharmacological stimulus such caffeine or cocaine

ndashBehaviorenvironment changes such as sleep deprivation pain

Chen Lin PhD

Xu G et al Neurology 2007

CBF changes of congnitively normal (CN) amp amnestic mild cognitive impairment (aMCI) groups

A significant regional CBF percent increase occurred in the region of the right parahippocampus during the sustained memory-encoding period in the CN group but not in the aMCI group

Chen Lin PhD

ASL versus BOLD for fMRIbull Much lower SNR

1-2 signal change in ASL versus 5-10 in BOLD

bull Direct measure of CBF changesBOLD is coupled with CBF CMRO2 CBV

bull More localized and quantitative

bull Better long-term stability and reproducibilityLow frequency fMRI

bull Insensitive to magnetic susceptibility effectsUse short TE and susceptibility insensitive image acquisition

bull Less inter-subject variabilityAllows group analysis long term and multisite study

Chen Lin PhD

ASL versus DSC

bull Lower SNR and therefore typical acquired with lower resolution

bull More susceptible to physiological variation

bull Less coverage

bull Non-invasive

Chen Lin PhD

7142009

8

ASL Outside of the Brain

bull Kidney (Warmuth JMRI 2007)

bull Muscle (Frouin MRM 2006)

bull Lung (Bolar MRM 2006)

Chen Lin PhD

Getting more from Diffusion Attenuation Curve

Ln(S)

b

Perfusion

S = Sb=0 [ (1-f) endashbD + f endashb(D+D) ]

b = 50

b = 500

Chen Lin PhD

Intra Voxel Incoherence Imaging

Luciani A et al Radiology 2008 Dec249(3)891-9

DWI b=50 D

D fChen Lin PhD

Thank you

Page 3: Chen Lin, PhD - Indiana University Bloomingtonmri/CE/slides/MR Perfusion... · Chen Lin, PhD DSC Post-processing (Siemens) 1. Define a ROI for selecting AIF 2. Select an AIF among

7142009

3

Tumor versus Radiation Necrosis

T2w rCBV

Chen Lin PhD

Leakage and Permeability

Chen Lin PhD

Typical DCE Perfusion Protocol

bull Spoiled 2D or 3D gradient echo (SPGR or FLASH)

bull TE = min

bull TR = min ( typically 4 ndash 10 sec)

bull FA = 30 ndash 40 deg

bull TA = several minutes

bull Contrast dose = 01 ndash 02 mmolkg

bull Injection rate 3 ndash 5 mlsec with 20 ml saline flush

bull Measure baseline T1 with different FAsChen Lin PhD

DCE-MRI Evaluation

Qualitative - shape of signal intensity (SI) data curve

Semi-quantitative - indices that describe one or more parts of SI or [Gd] curves

Upslope gradient max amplitude washout rate or area under curve at a fixed time point

True quantitative - indices from contrast medium concentration changes using pharmacokinetic modelling

Chen Lin PhD

Type I(semi-necrotic with reactive changes)

Type II(viable tumor)

Type III(rapidly proliferating

tumor edge)

(Taylor and Reddick Adv Drug Del Rev 2000)

Qualitative Evaluation of DCE-MRI time Curve

Chen Lin PhD

DCE Perfusion Modelling

( )( ) ( )transt

p ep t

trans

ep

dC tK C t k C t

dt

Kk

Ve

Chen Lin PhD

Blood plasma

Cp

Ct = Ce Ve

Ve Ce

Ktrans

Kep

Tumor Extracellular

Space

NormalExtracellular

Space

7142009

4

Quantification of Contrast Concentration

R1(t) = R1(0) + C(t) Rbull R1(t) Longitudinal relaxation rate 1T1(t) at a given time

point t can be determined from the ratio of signal intensities between T1w and PDw images SI(T1w)SI(PDw)

bull R1(0) Longitudinal relaxation rate 1T1(0) at baseline ie before contrast arrival can also be determined from SI(T1w)SI(PDw) at baseline

bull R Relaxivity of the contrast agent ie Gd-DTPA

bull C(t) Contrast concentration at a given time t can be derived

bull Ct(t) Tissue contrast concentration curve

bull Cp(t) Blood contrast concentration curve arterial input function (AIF)

Chen Lin PhD

Quantitative analysis with pharmacokinetic modelling

bull Advantagesndash Whole curve shape is analysed

ndash Biologically relevant physiological parameters

ndash Independent of scanner strength manufacturer and imaging routines

ndash Enables valid comparisons of serial measurements and data exchange between different imaging centres

bull Disadvantagesndash Data acquisition and analysis is more complex

ndash Lack of commercial software for analysis

ndash Models may not fit the data observed

Chen Lin PhD

DCE Perfusion of Pancreatic Tumor

Chen Lin PhD

DCE Perfusion of Pancreatic Tumor

Chen Lin PhD

PWI with Arterial Spinning Labeling

bull Tagged blood flows through vasculature and exchanges with tissue

bull Change total tissue magnetization in slice (1-2) compared to control

bull ldquoControlrdquo ndash ldquoLabeledrdquo CBF

Artery Vein

Chen Lin PhD

Labeling Techniques

bull Continuous or Steady-state ASL (CASL)

ndash Saturation (Detre 1992)

ndash Inversion (Williams 1992) with flow-driven adiabatic inversion pulse

ndash Labeling with 2nd coil

bull Pulsed ASL (uses a bolus of labeled spins) (PASL)

ndash STAR TILT PICORE FAIR hellip

bull Velocity Selective ASL

Chen Lin PhD

7142009

5

CASL with Dedicated Labeling Coil

bull Reduce SAR

bull Reduce MT effect

bull Additional hardware

Imaging Plane

Labeling Coil

Chen Lin PhD

Typical PASL Design

M0 Label Control Label ControlLabelControl

Control - Label Control - Label Control - LabelDM

CBF = 3x108 l [mlg] a [] TI1 Exp(TI2T1blood) DMM0

Image

Acquisition

Module

TI2

TI1Labeling

Control

Module

Bolus Cut-off Module

Chen Lin PhD

Chen Lin PhD

QUIPSS amp QUIPSS II

bull QUantitative Imaging of Perfusion using a Single Subtraction

bull Use saturation pulses to shave the leading or trailing edge of the bolus so the transit time is the same

Image

Acquisition

Module

TI2

TI1Labeling

Control

Module

900 900

ASL Image Acquisition Mode

bull 2D single and multiple slices

bull 3D (long acquisition time)

Chen Lin PhD

ASL Image Acquisition Methods

bull GRESE EPI (STAR + EPI Edelman 1994 FAIR + EPI Kim 1995)

bull HASTE (STAR + HAlf-fourier Single-shot Turbo spin Echo Chen 1997)

bull GRASE (FAIR + GRadient echo And Spin Echo Crelier 1999)

bull TrueFISP (FAIR + True Fast Imaging with Steady-state Precession Boss 2007)

Chen Lin PhD

PASL Issues and Optimization

bull SNR and Susceptibility artifacts

ndash Use short TE amp ETL

bull T1 Relaxation effect amp Transit Time Difference

ndash Decay of magnetization from the tagging location to imaging location

ndash The transit time of leading and trailing edge of the labeled bolus are different

bull Arterial Signal

ndash Apply a small bipolar diffusion weighting gradient b ~ 2 smm2

Chen Lin PhD

7142009

6

PASL Issues and Optimization

bull Physiological Noise

ndash Reduction with background suppression

ndash Model and correct the physiological fluctuation with statistical analysis (GLM)

bull MT effect

ndash Suppression at imaging location from off-resonance effect of labeling pulse Asymmetric with respect to frequency shift

bull Eddy Current Compensation

ndash Maintain the similar gradient waveform between tagging and control

Chen Lin PhD

MT Effect amp Labeling Schemes

bull STAR (Signal Targeting with Alternating Radio Frequency)

bull PICORE (Proximal Inversion with a Control for Off-resonance Effect) (Wong 1997)

bull TILT (Transfer Insensitive Labeling Technique) (Gobay 1999)

bull FAIR (Flow-sensitive Alternating Inversion Recovery) (Kim 1995)

Chen Lin PhD

PICORE

Tagging

Control

1800

1800

wt = wi + Dw

wc = wi + Dw

Gt = 0

Chen Lin PhD

FAIR

Tagging

Control

1800

1800

wt = wi

wc = wi

Gt = 0Chen Lin PhD

ASL Applications

bull Quantification of rest state CBF and study the perfusion changes from a disease or treatmentintervention

ndash Tumor Stroke Alzheimerrsquos Disease vascular disorder etc hellip

ndash Infant brain development (Wang Neuroimage 2008) amp Aging (Biagi JMRI 2007)

ndash Uncoupling contributions to BOLD fMRI

Chen Lin PhD

AVM

T2wDSCrMTT

ASLrCBF

DSCNEI

Chen Lin PhD

7142009

7

T2-weighted

Ischemia (RgtL)

Pre-op ASL

Hypo-perfusion (RgtL)

Post-op ASL

Bilateral increase

ASL perfusion correlated better with cognitive deficits than structural MRI

Jefferson AJNR 2006

Intracranial Stenosis

Chen Lin PhD

Du A T et al Neurology 200667125-1220

Frontotemporal Dementia (FTD) and Alzheimer Disease (AD)

FTD and AD display different spatial distributions of hypo-perfusion Chen Lin PhD

ASL Perfusion based fMRI

bull Perfusion fMRI ie Quantification of functional related dynamic CBF changes

ndashConventional stimulustasks such as sensory-motor memory speech tasks similar to those in BOLD fMRI

ndashPharmacological stimulus such caffeine or cocaine

ndashBehaviorenvironment changes such as sleep deprivation pain

Chen Lin PhD

Xu G et al Neurology 2007

CBF changes of congnitively normal (CN) amp amnestic mild cognitive impairment (aMCI) groups

A significant regional CBF percent increase occurred in the region of the right parahippocampus during the sustained memory-encoding period in the CN group but not in the aMCI group

Chen Lin PhD

ASL versus BOLD for fMRIbull Much lower SNR

1-2 signal change in ASL versus 5-10 in BOLD

bull Direct measure of CBF changesBOLD is coupled with CBF CMRO2 CBV

bull More localized and quantitative

bull Better long-term stability and reproducibilityLow frequency fMRI

bull Insensitive to magnetic susceptibility effectsUse short TE and susceptibility insensitive image acquisition

bull Less inter-subject variabilityAllows group analysis long term and multisite study

Chen Lin PhD

ASL versus DSC

bull Lower SNR and therefore typical acquired with lower resolution

bull More susceptible to physiological variation

bull Less coverage

bull Non-invasive

Chen Lin PhD

7142009

8

ASL Outside of the Brain

bull Kidney (Warmuth JMRI 2007)

bull Muscle (Frouin MRM 2006)

bull Lung (Bolar MRM 2006)

Chen Lin PhD

Getting more from Diffusion Attenuation Curve

Ln(S)

b

Perfusion

S = Sb=0 [ (1-f) endashbD + f endashb(D+D) ]

b = 50

b = 500

Chen Lin PhD

Intra Voxel Incoherence Imaging

Luciani A et al Radiology 2008 Dec249(3)891-9

DWI b=50 D

D fChen Lin PhD

Thank you

Page 4: Chen Lin, PhD - Indiana University Bloomingtonmri/CE/slides/MR Perfusion... · Chen Lin, PhD DSC Post-processing (Siemens) 1. Define a ROI for selecting AIF 2. Select an AIF among

7142009

4

Quantification of Contrast Concentration

R1(t) = R1(0) + C(t) Rbull R1(t) Longitudinal relaxation rate 1T1(t) at a given time

point t can be determined from the ratio of signal intensities between T1w and PDw images SI(T1w)SI(PDw)

bull R1(0) Longitudinal relaxation rate 1T1(0) at baseline ie before contrast arrival can also be determined from SI(T1w)SI(PDw) at baseline

bull R Relaxivity of the contrast agent ie Gd-DTPA

bull C(t) Contrast concentration at a given time t can be derived

bull Ct(t) Tissue contrast concentration curve

bull Cp(t) Blood contrast concentration curve arterial input function (AIF)

Chen Lin PhD

Quantitative analysis with pharmacokinetic modelling

bull Advantagesndash Whole curve shape is analysed

ndash Biologically relevant physiological parameters

ndash Independent of scanner strength manufacturer and imaging routines

ndash Enables valid comparisons of serial measurements and data exchange between different imaging centres

bull Disadvantagesndash Data acquisition and analysis is more complex

ndash Lack of commercial software for analysis

ndash Models may not fit the data observed

Chen Lin PhD

DCE Perfusion of Pancreatic Tumor

Chen Lin PhD

DCE Perfusion of Pancreatic Tumor

Chen Lin PhD

PWI with Arterial Spinning Labeling

bull Tagged blood flows through vasculature and exchanges with tissue

bull Change total tissue magnetization in slice (1-2) compared to control

bull ldquoControlrdquo ndash ldquoLabeledrdquo CBF

Artery Vein

Chen Lin PhD

Labeling Techniques

bull Continuous or Steady-state ASL (CASL)

ndash Saturation (Detre 1992)

ndash Inversion (Williams 1992) with flow-driven adiabatic inversion pulse

ndash Labeling with 2nd coil

bull Pulsed ASL (uses a bolus of labeled spins) (PASL)

ndash STAR TILT PICORE FAIR hellip

bull Velocity Selective ASL

Chen Lin PhD

7142009

5

CASL with Dedicated Labeling Coil

bull Reduce SAR

bull Reduce MT effect

bull Additional hardware

Imaging Plane

Labeling Coil

Chen Lin PhD

Typical PASL Design

M0 Label Control Label ControlLabelControl

Control - Label Control - Label Control - LabelDM

CBF = 3x108 l [mlg] a [] TI1 Exp(TI2T1blood) DMM0

Image

Acquisition

Module

TI2

TI1Labeling

Control

Module

Bolus Cut-off Module

Chen Lin PhD

Chen Lin PhD

QUIPSS amp QUIPSS II

bull QUantitative Imaging of Perfusion using a Single Subtraction

bull Use saturation pulses to shave the leading or trailing edge of the bolus so the transit time is the same

Image

Acquisition

Module

TI2

TI1Labeling

Control

Module

900 900

ASL Image Acquisition Mode

bull 2D single and multiple slices

bull 3D (long acquisition time)

Chen Lin PhD

ASL Image Acquisition Methods

bull GRESE EPI (STAR + EPI Edelman 1994 FAIR + EPI Kim 1995)

bull HASTE (STAR + HAlf-fourier Single-shot Turbo spin Echo Chen 1997)

bull GRASE (FAIR + GRadient echo And Spin Echo Crelier 1999)

bull TrueFISP (FAIR + True Fast Imaging with Steady-state Precession Boss 2007)

Chen Lin PhD

PASL Issues and Optimization

bull SNR and Susceptibility artifacts

ndash Use short TE amp ETL

bull T1 Relaxation effect amp Transit Time Difference

ndash Decay of magnetization from the tagging location to imaging location

ndash The transit time of leading and trailing edge of the labeled bolus are different

bull Arterial Signal

ndash Apply a small bipolar diffusion weighting gradient b ~ 2 smm2

Chen Lin PhD

7142009

6

PASL Issues and Optimization

bull Physiological Noise

ndash Reduction with background suppression

ndash Model and correct the physiological fluctuation with statistical analysis (GLM)

bull MT effect

ndash Suppression at imaging location from off-resonance effect of labeling pulse Asymmetric with respect to frequency shift

bull Eddy Current Compensation

ndash Maintain the similar gradient waveform between tagging and control

Chen Lin PhD

MT Effect amp Labeling Schemes

bull STAR (Signal Targeting with Alternating Radio Frequency)

bull PICORE (Proximal Inversion with a Control for Off-resonance Effect) (Wong 1997)

bull TILT (Transfer Insensitive Labeling Technique) (Gobay 1999)

bull FAIR (Flow-sensitive Alternating Inversion Recovery) (Kim 1995)

Chen Lin PhD

PICORE

Tagging

Control

1800

1800

wt = wi + Dw

wc = wi + Dw

Gt = 0

Chen Lin PhD

FAIR

Tagging

Control

1800

1800

wt = wi

wc = wi

Gt = 0Chen Lin PhD

ASL Applications

bull Quantification of rest state CBF and study the perfusion changes from a disease or treatmentintervention

ndash Tumor Stroke Alzheimerrsquos Disease vascular disorder etc hellip

ndash Infant brain development (Wang Neuroimage 2008) amp Aging (Biagi JMRI 2007)

ndash Uncoupling contributions to BOLD fMRI

Chen Lin PhD

AVM

T2wDSCrMTT

ASLrCBF

DSCNEI

Chen Lin PhD

7142009

7

T2-weighted

Ischemia (RgtL)

Pre-op ASL

Hypo-perfusion (RgtL)

Post-op ASL

Bilateral increase

ASL perfusion correlated better with cognitive deficits than structural MRI

Jefferson AJNR 2006

Intracranial Stenosis

Chen Lin PhD

Du A T et al Neurology 200667125-1220

Frontotemporal Dementia (FTD) and Alzheimer Disease (AD)

FTD and AD display different spatial distributions of hypo-perfusion Chen Lin PhD

ASL Perfusion based fMRI

bull Perfusion fMRI ie Quantification of functional related dynamic CBF changes

ndashConventional stimulustasks such as sensory-motor memory speech tasks similar to those in BOLD fMRI

ndashPharmacological stimulus such caffeine or cocaine

ndashBehaviorenvironment changes such as sleep deprivation pain

Chen Lin PhD

Xu G et al Neurology 2007

CBF changes of congnitively normal (CN) amp amnestic mild cognitive impairment (aMCI) groups

A significant regional CBF percent increase occurred in the region of the right parahippocampus during the sustained memory-encoding period in the CN group but not in the aMCI group

Chen Lin PhD

ASL versus BOLD for fMRIbull Much lower SNR

1-2 signal change in ASL versus 5-10 in BOLD

bull Direct measure of CBF changesBOLD is coupled with CBF CMRO2 CBV

bull More localized and quantitative

bull Better long-term stability and reproducibilityLow frequency fMRI

bull Insensitive to magnetic susceptibility effectsUse short TE and susceptibility insensitive image acquisition

bull Less inter-subject variabilityAllows group analysis long term and multisite study

Chen Lin PhD

ASL versus DSC

bull Lower SNR and therefore typical acquired with lower resolution

bull More susceptible to physiological variation

bull Less coverage

bull Non-invasive

Chen Lin PhD

7142009

8

ASL Outside of the Brain

bull Kidney (Warmuth JMRI 2007)

bull Muscle (Frouin MRM 2006)

bull Lung (Bolar MRM 2006)

Chen Lin PhD

Getting more from Diffusion Attenuation Curve

Ln(S)

b

Perfusion

S = Sb=0 [ (1-f) endashbD + f endashb(D+D) ]

b = 50

b = 500

Chen Lin PhD

Intra Voxel Incoherence Imaging

Luciani A et al Radiology 2008 Dec249(3)891-9

DWI b=50 D

D fChen Lin PhD

Thank you

Page 5: Chen Lin, PhD - Indiana University Bloomingtonmri/CE/slides/MR Perfusion... · Chen Lin, PhD DSC Post-processing (Siemens) 1. Define a ROI for selecting AIF 2. Select an AIF among

7142009

5

CASL with Dedicated Labeling Coil

bull Reduce SAR

bull Reduce MT effect

bull Additional hardware

Imaging Plane

Labeling Coil

Chen Lin PhD

Typical PASL Design

M0 Label Control Label ControlLabelControl

Control - Label Control - Label Control - LabelDM

CBF = 3x108 l [mlg] a [] TI1 Exp(TI2T1blood) DMM0

Image

Acquisition

Module

TI2

TI1Labeling

Control

Module

Bolus Cut-off Module

Chen Lin PhD

Chen Lin PhD

QUIPSS amp QUIPSS II

bull QUantitative Imaging of Perfusion using a Single Subtraction

bull Use saturation pulses to shave the leading or trailing edge of the bolus so the transit time is the same

Image

Acquisition

Module

TI2

TI1Labeling

Control

Module

900 900

ASL Image Acquisition Mode

bull 2D single and multiple slices

bull 3D (long acquisition time)

Chen Lin PhD

ASL Image Acquisition Methods

bull GRESE EPI (STAR + EPI Edelman 1994 FAIR + EPI Kim 1995)

bull HASTE (STAR + HAlf-fourier Single-shot Turbo spin Echo Chen 1997)

bull GRASE (FAIR + GRadient echo And Spin Echo Crelier 1999)

bull TrueFISP (FAIR + True Fast Imaging with Steady-state Precession Boss 2007)

Chen Lin PhD

PASL Issues and Optimization

bull SNR and Susceptibility artifacts

ndash Use short TE amp ETL

bull T1 Relaxation effect amp Transit Time Difference

ndash Decay of magnetization from the tagging location to imaging location

ndash The transit time of leading and trailing edge of the labeled bolus are different

bull Arterial Signal

ndash Apply a small bipolar diffusion weighting gradient b ~ 2 smm2

Chen Lin PhD

7142009

6

PASL Issues and Optimization

bull Physiological Noise

ndash Reduction with background suppression

ndash Model and correct the physiological fluctuation with statistical analysis (GLM)

bull MT effect

ndash Suppression at imaging location from off-resonance effect of labeling pulse Asymmetric with respect to frequency shift

bull Eddy Current Compensation

ndash Maintain the similar gradient waveform between tagging and control

Chen Lin PhD

MT Effect amp Labeling Schemes

bull STAR (Signal Targeting with Alternating Radio Frequency)

bull PICORE (Proximal Inversion with a Control for Off-resonance Effect) (Wong 1997)

bull TILT (Transfer Insensitive Labeling Technique) (Gobay 1999)

bull FAIR (Flow-sensitive Alternating Inversion Recovery) (Kim 1995)

Chen Lin PhD

PICORE

Tagging

Control

1800

1800

wt = wi + Dw

wc = wi + Dw

Gt = 0

Chen Lin PhD

FAIR

Tagging

Control

1800

1800

wt = wi

wc = wi

Gt = 0Chen Lin PhD

ASL Applications

bull Quantification of rest state CBF and study the perfusion changes from a disease or treatmentintervention

ndash Tumor Stroke Alzheimerrsquos Disease vascular disorder etc hellip

ndash Infant brain development (Wang Neuroimage 2008) amp Aging (Biagi JMRI 2007)

ndash Uncoupling contributions to BOLD fMRI

Chen Lin PhD

AVM

T2wDSCrMTT

ASLrCBF

DSCNEI

Chen Lin PhD

7142009

7

T2-weighted

Ischemia (RgtL)

Pre-op ASL

Hypo-perfusion (RgtL)

Post-op ASL

Bilateral increase

ASL perfusion correlated better with cognitive deficits than structural MRI

Jefferson AJNR 2006

Intracranial Stenosis

Chen Lin PhD

Du A T et al Neurology 200667125-1220

Frontotemporal Dementia (FTD) and Alzheimer Disease (AD)

FTD and AD display different spatial distributions of hypo-perfusion Chen Lin PhD

ASL Perfusion based fMRI

bull Perfusion fMRI ie Quantification of functional related dynamic CBF changes

ndashConventional stimulustasks such as sensory-motor memory speech tasks similar to those in BOLD fMRI

ndashPharmacological stimulus such caffeine or cocaine

ndashBehaviorenvironment changes such as sleep deprivation pain

Chen Lin PhD

Xu G et al Neurology 2007

CBF changes of congnitively normal (CN) amp amnestic mild cognitive impairment (aMCI) groups

A significant regional CBF percent increase occurred in the region of the right parahippocampus during the sustained memory-encoding period in the CN group but not in the aMCI group

Chen Lin PhD

ASL versus BOLD for fMRIbull Much lower SNR

1-2 signal change in ASL versus 5-10 in BOLD

bull Direct measure of CBF changesBOLD is coupled with CBF CMRO2 CBV

bull More localized and quantitative

bull Better long-term stability and reproducibilityLow frequency fMRI

bull Insensitive to magnetic susceptibility effectsUse short TE and susceptibility insensitive image acquisition

bull Less inter-subject variabilityAllows group analysis long term and multisite study

Chen Lin PhD

ASL versus DSC

bull Lower SNR and therefore typical acquired with lower resolution

bull More susceptible to physiological variation

bull Less coverage

bull Non-invasive

Chen Lin PhD

7142009

8

ASL Outside of the Brain

bull Kidney (Warmuth JMRI 2007)

bull Muscle (Frouin MRM 2006)

bull Lung (Bolar MRM 2006)

Chen Lin PhD

Getting more from Diffusion Attenuation Curve

Ln(S)

b

Perfusion

S = Sb=0 [ (1-f) endashbD + f endashb(D+D) ]

b = 50

b = 500

Chen Lin PhD

Intra Voxel Incoherence Imaging

Luciani A et al Radiology 2008 Dec249(3)891-9

DWI b=50 D

D fChen Lin PhD

Thank you

Page 6: Chen Lin, PhD - Indiana University Bloomingtonmri/CE/slides/MR Perfusion... · Chen Lin, PhD DSC Post-processing (Siemens) 1. Define a ROI for selecting AIF 2. Select an AIF among

7142009

6

PASL Issues and Optimization

bull Physiological Noise

ndash Reduction with background suppression

ndash Model and correct the physiological fluctuation with statistical analysis (GLM)

bull MT effect

ndash Suppression at imaging location from off-resonance effect of labeling pulse Asymmetric with respect to frequency shift

bull Eddy Current Compensation

ndash Maintain the similar gradient waveform between tagging and control

Chen Lin PhD

MT Effect amp Labeling Schemes

bull STAR (Signal Targeting with Alternating Radio Frequency)

bull PICORE (Proximal Inversion with a Control for Off-resonance Effect) (Wong 1997)

bull TILT (Transfer Insensitive Labeling Technique) (Gobay 1999)

bull FAIR (Flow-sensitive Alternating Inversion Recovery) (Kim 1995)

Chen Lin PhD

PICORE

Tagging

Control

1800

1800

wt = wi + Dw

wc = wi + Dw

Gt = 0

Chen Lin PhD

FAIR

Tagging

Control

1800

1800

wt = wi

wc = wi

Gt = 0Chen Lin PhD

ASL Applications

bull Quantification of rest state CBF and study the perfusion changes from a disease or treatmentintervention

ndash Tumor Stroke Alzheimerrsquos Disease vascular disorder etc hellip

ndash Infant brain development (Wang Neuroimage 2008) amp Aging (Biagi JMRI 2007)

ndash Uncoupling contributions to BOLD fMRI

Chen Lin PhD

AVM

T2wDSCrMTT

ASLrCBF

DSCNEI

Chen Lin PhD

7142009

7

T2-weighted

Ischemia (RgtL)

Pre-op ASL

Hypo-perfusion (RgtL)

Post-op ASL

Bilateral increase

ASL perfusion correlated better with cognitive deficits than structural MRI

Jefferson AJNR 2006

Intracranial Stenosis

Chen Lin PhD

Du A T et al Neurology 200667125-1220

Frontotemporal Dementia (FTD) and Alzheimer Disease (AD)

FTD and AD display different spatial distributions of hypo-perfusion Chen Lin PhD

ASL Perfusion based fMRI

bull Perfusion fMRI ie Quantification of functional related dynamic CBF changes

ndashConventional stimulustasks such as sensory-motor memory speech tasks similar to those in BOLD fMRI

ndashPharmacological stimulus such caffeine or cocaine

ndashBehaviorenvironment changes such as sleep deprivation pain

Chen Lin PhD

Xu G et al Neurology 2007

CBF changes of congnitively normal (CN) amp amnestic mild cognitive impairment (aMCI) groups

A significant regional CBF percent increase occurred in the region of the right parahippocampus during the sustained memory-encoding period in the CN group but not in the aMCI group

Chen Lin PhD

ASL versus BOLD for fMRIbull Much lower SNR

1-2 signal change in ASL versus 5-10 in BOLD

bull Direct measure of CBF changesBOLD is coupled with CBF CMRO2 CBV

bull More localized and quantitative

bull Better long-term stability and reproducibilityLow frequency fMRI

bull Insensitive to magnetic susceptibility effectsUse short TE and susceptibility insensitive image acquisition

bull Less inter-subject variabilityAllows group analysis long term and multisite study

Chen Lin PhD

ASL versus DSC

bull Lower SNR and therefore typical acquired with lower resolution

bull More susceptible to physiological variation

bull Less coverage

bull Non-invasive

Chen Lin PhD

7142009

8

ASL Outside of the Brain

bull Kidney (Warmuth JMRI 2007)

bull Muscle (Frouin MRM 2006)

bull Lung (Bolar MRM 2006)

Chen Lin PhD

Getting more from Diffusion Attenuation Curve

Ln(S)

b

Perfusion

S = Sb=0 [ (1-f) endashbD + f endashb(D+D) ]

b = 50

b = 500

Chen Lin PhD

Intra Voxel Incoherence Imaging

Luciani A et al Radiology 2008 Dec249(3)891-9

DWI b=50 D

D fChen Lin PhD

Thank you

Page 7: Chen Lin, PhD - Indiana University Bloomingtonmri/CE/slides/MR Perfusion... · Chen Lin, PhD DSC Post-processing (Siemens) 1. Define a ROI for selecting AIF 2. Select an AIF among

7142009

7

T2-weighted

Ischemia (RgtL)

Pre-op ASL

Hypo-perfusion (RgtL)

Post-op ASL

Bilateral increase

ASL perfusion correlated better with cognitive deficits than structural MRI

Jefferson AJNR 2006

Intracranial Stenosis

Chen Lin PhD

Du A T et al Neurology 200667125-1220

Frontotemporal Dementia (FTD) and Alzheimer Disease (AD)

FTD and AD display different spatial distributions of hypo-perfusion Chen Lin PhD

ASL Perfusion based fMRI

bull Perfusion fMRI ie Quantification of functional related dynamic CBF changes

ndashConventional stimulustasks such as sensory-motor memory speech tasks similar to those in BOLD fMRI

ndashPharmacological stimulus such caffeine or cocaine

ndashBehaviorenvironment changes such as sleep deprivation pain

Chen Lin PhD

Xu G et al Neurology 2007

CBF changes of congnitively normal (CN) amp amnestic mild cognitive impairment (aMCI) groups

A significant regional CBF percent increase occurred in the region of the right parahippocampus during the sustained memory-encoding period in the CN group but not in the aMCI group

Chen Lin PhD

ASL versus BOLD for fMRIbull Much lower SNR

1-2 signal change in ASL versus 5-10 in BOLD

bull Direct measure of CBF changesBOLD is coupled with CBF CMRO2 CBV

bull More localized and quantitative

bull Better long-term stability and reproducibilityLow frequency fMRI

bull Insensitive to magnetic susceptibility effectsUse short TE and susceptibility insensitive image acquisition

bull Less inter-subject variabilityAllows group analysis long term and multisite study

Chen Lin PhD

ASL versus DSC

bull Lower SNR and therefore typical acquired with lower resolution

bull More susceptible to physiological variation

bull Less coverage

bull Non-invasive

Chen Lin PhD

7142009

8

ASL Outside of the Brain

bull Kidney (Warmuth JMRI 2007)

bull Muscle (Frouin MRM 2006)

bull Lung (Bolar MRM 2006)

Chen Lin PhD

Getting more from Diffusion Attenuation Curve

Ln(S)

b

Perfusion

S = Sb=0 [ (1-f) endashbD + f endashb(D+D) ]

b = 50

b = 500

Chen Lin PhD

Intra Voxel Incoherence Imaging

Luciani A et al Radiology 2008 Dec249(3)891-9

DWI b=50 D

D fChen Lin PhD

Thank you

Page 8: Chen Lin, PhD - Indiana University Bloomingtonmri/CE/slides/MR Perfusion... · Chen Lin, PhD DSC Post-processing (Siemens) 1. Define a ROI for selecting AIF 2. Select an AIF among

7142009

8

ASL Outside of the Brain

bull Kidney (Warmuth JMRI 2007)

bull Muscle (Frouin MRM 2006)

bull Lung (Bolar MRM 2006)

Chen Lin PhD

Getting more from Diffusion Attenuation Curve

Ln(S)

b

Perfusion

S = Sb=0 [ (1-f) endashbD + f endashb(D+D) ]

b = 50

b = 500

Chen Lin PhD

Intra Voxel Incoherence Imaging

Luciani A et al Radiology 2008 Dec249(3)891-9

DWI b=50 D

D fChen Lin PhD

Thank you