C a - Rigshospitalet

Transcription

C a - Rigshospitalet
Basic perfusion theory
January 24th 2012
by
Henrik BW Larsson
Functional Imaging Unit, Diagnostic Department
HBWL
Outline
• 
• 
• 
• 
• 
What is perfusion
Why measure perfusion
Measures
The easy part: What to do and why
The complicated part: Formalism and
models
•  The really complicated part:
Deconvolution
•  Conclusion
HBWL
What is perfusion?
Large vessels : flow
½ mm
Perfusion: related to the
microvascular system ~ the
capillaries
HBWL
The vascular system of the brain
and perfusion
Venules: capacity
vessels
20% Artery:conductance
vessels
50%
Veins
30% Capillaries: exchange
vessels ~
transport~diffusion
Arteriole:resistance
vessels
HBWL
Perfusion : ml/100g/min
½ mm
HBWL
Perfusion : ml/100g/min
½ mm
HBWL
Perfusion metrics in imaging: ml/min/
100g or ml/min/100ml
½ mm
HBWL
Number of transport (ml) vehicles entering
100 ml tissue pr. time unit::
20 - 80 ml/min/100 ml tissue volume
HBWL
Important metrics
•  Perfusion: – f [ml/min/100g] or [ ml/min/100ml ]
•  Brain Perfusion (‘flow’) : Cerebral blood flow CBF [ml/100g/min]
•  Cerebral blood volume: CBV [ml/100g]
½ mm
•  Volume of Distibution: Vd [ml/100g] or [ml/100ml]
•  Mean transit time: MTT [s]
•  Blood brain permeability: PS product [ml/100g/min]
HBWL
Important metrics
•  Perfusion: – f [ml/min/100g] or [ ml/min/100ml ]
•  Brain Perfusion (‘flow’) : Cerebral blood flow CBF [ml/100g/min]
•  Cerebral blood volume: CBV [ml/100g]
½ mm
•  Mean transit time: MTT [s]
•  Blood brain permeability: PS product [ml/100g/min]
HBWL
Blood flow changes and energy metabolism in
brain and skeletal muscle
Brain
Muscle
5 – 30 %
Activation
Increases up
to 30 x
Rest
HBWL
Why measure brain perfusion?
•  It intimately related to brain
activation
•  Govern oxygen delivery and CMRO2:
CMRO2=CBF x (Ca – Cv) = CBF x Ca x OEF
•  Is profoundly changed in nearly all
brain diseases either primarily or
secondary or in a more subtle way
HBWL
HBWL
Non-invasive perfusion:
What to do
and
the easy part
HBWL
Measuring perfusion by an external
registration: CT,SPECT,PET,MRI
detector
artery
vein
f
f: perfusion in [ml/min /100g]
HBWL
How can it be measured ?
Add a contrast agent carried by the blood to the tissue
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
Bolus of
tracer or
contrast
HBWL
How can it be measured ?
Add a contrast agent carried by the blood to the tissue
Contrast agent
exogent
endogent
HBWL
The complicated part:
Single bolus injection
and
external registration
HBWL
The complicated part:
Single bolus injection
and
external registration
HBWL
time
HBWL
Signal ≈ Ctis(0) : f Ca(0)
Perfusion (f) = vehicles/min
ml/min
x
Conc (Ca) = the cargo they
carry
mmol/ml
time
HBWL
Ctis(t) = f Ca(0) Δt RF(t)
RF(t) =1 for t < MTT
Ctis(0) =: f Ca(0) Δt
RF(t) =0 for t > MTT
flux
MTT
time
dose
HBWL
Ctis(t) = f Ca(0) Δt RF(t)
Ctis(0) = f Ca(0) Δt
time
HBWL
Summing up: direct short bolus
Measure the
tissue conc
Measure the input conc i.e.
input function
Scanner signal : Ctis(t) = f Ca(0) Δt RF(t)
Estimate f and RF(t)
HBWL
Summing up
Measure the
tissue conc
Measure the input conc i.e.
input function
Scanner signal : Ctis(t) = f Ca(0) Δt RF(t)
Estimate f and RF(t)
RF(t) by ’model free’ methods, or assume a model
e.g. RF(t) =e-k2t
HBWL
Different perfusion tracers
behaves differently
HBWL
Ctis(t) = f Ca(0) RF(t)
Ctis(0) = f Ca(0) Δt
time (s)
HBWL
Ctis(t) = f Ca(0) Δt RF(t)
RF(t) = e-k2t
Ctis(0) = f Ca(0) Δt
1
2
time (s)
HBWL
Ctis(t) = f Ca(0) Δt RF(t)
RF(t) = e-k2t
Ctis(0) = f Ca(0) Δt
100
200
time (s)
HBWL
Ctis(t) = f Ca(0) RF(t)
RF(t) = e-k2t
Ctis(0) = f Ca(0) Δt
100
200
time (s)
HBWL
Ctis(t) = f Ca(0) Δt RF(t)
RF(t) = e-k2t + e-k3t
Ctis(0) = f Ca(0) Δt
100
200
time (s)
HBWL
The residue impulse response function RF(t)
RF(t) : the fraction of the injected dose
remaining in the tissue (voxel) as a function of
time
Mean transit time : MTT
MTT = ∫RF(t)
0
HBWL
Mean transit time : MTT
MTT = ∫RF(t)dt
RF(t)
0
MTT=
RF(t)
1
t
∑
n
t
N
RF(t)
1
t
MTT
MTT
1
t
n
N
MTT
HBWL
Generally
Perfusion: f
Distribution vol: Vd
f=
Mean transit time: MTT
Vd
MTT
For an intravascular contrast agent, the case in brain
MRI we have:
Brain perfusion: CBF
Brain blood volume: CBV
CBF =
CBV
MTT
Mean transit time: MTT
HBWL
Generally
For an intravascular contrast agent, the case in brain
perfusion MRI we have:
Brain perfusion: CBF
Brain blood volume: CBV
Mean transit time: MTT
CBF =
CBV
MTT
HBWL
The really complicated part:
Deconvolution
HBWL
We cannot apply a bolus directly in the tissue !
HBWL
Input :
Tissue enhancement :
∞
Ca(t)
Ctis(t) = ∫f Ca(τ) RF(t- τ) dτ
0
tissue
Deconvolution :
find f RF(t)
f
time
HBWL
Input : Ca(t)
Tissue enhancement :
Ctis(t) = ?
tissue
HBWL
Input : Ca(t)
Tissue enhancement :
Ctis(t) = f Ca(0) RF(t) Δt
tissue
HBWL
Input : Ca(t)
Tissue enhancement :
Ctis(t) = ?
tissue
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = ?
tissue
If the linearity of the system exist
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(0) RF(t - 0) Δτ
tissue
0
0
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(1) RF(t - 1) Δτ
tissue
1
1
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(2) RF(t - 2) Δτ
tissue
2
2
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(3) RF(t - 3) Δτ
tissue
3
3
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(4) RF(t - 4) Δτ
tissue
4
4
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(5) RF(t - 5) Δτ
tissue
5
5
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(6) RF(t - 6) Δτ
tissue
6
6
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(7) RF(t - 7) Δτ
tissue
7
7
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(8) RF(t - 8) Δτ
tissue
8
8
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(9) RF(t - 9) Δτ
tissue
9
9
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(10) RF(t - 10) Δτ
tissue
10
10
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(11) RF(t - 11) Δτ
tissue
11
11
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(12) RF(t - 12) Δτ
tissue
12
12
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(13) RF(t - 13) Δτ
tissue
13
13
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(14) RF(t - 14) Δτ
tissue
14
14
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(15) RF(t - 15) Δτ
tissue
15
15
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(16) RF(t - 16) Δτ
tissue
16
16
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(17) RF(t - 17) Δτ
tissue
17
17
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(18) RF(t - 18) Δτ
tissue
18
18
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(19) RF(t - 19) Δτ
tissue
19
19
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(20) RF(t - 20) Δτ
tissue
20
20
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(21) RF(t - 21) Δτ
tissue
21
21
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(22) RF(t - 22) Δτ
tissue
22
22
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(23) RF(t - 23) Δτ
tissue
23
23
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(24) RF(t - 24) Δτ
tissue
24
24
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(25) RF(t - 25) Δτ
tissue
25
25
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(26) RF(t - 26) Δτ
tissue
26
26
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(27) RF(t - 27) Δτ
tissue
27
27
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(28) RF(t - 28) Δτ
tissue
28
28
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(29) RF(t - 29) Δτ
tissue
29
29
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(30) RF(t - 30) Δτ
tissue
30
30
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(31) RF(t - 31) Δτ
tissue
31
31
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(32) RF(t - 32) Δτ
tissue
32
32
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(32) RF(t - 32) Δτ
tissue
32
32
HBWL
Input : Ca(t)
Tissue enhancement :
composed of many
small input
Ctis(t) = f Ca(33) RF(t - 33) Δτ
tissue
33
33
HBWL
Tissue enhancement :
f Ca(τ) RF(t - τ ) Δτ ; τ = 0:33
τ: 0
20
33
HBWL
Total tissue enhancement :
Ctis(t)= ∑f Ca(τ) RF(t - τ ) Δτ ; τ = 0:33
τ: 0
20
33
HBWL
Total tissue concentration:
t
Ctis(t)= ∫f Ca(τ) RF(t - τ ) dτ
0
The convolution integral
τ: 0
20
33
HBWL
Input :
Tissue enhancement :
Ca(t)
Ctis(t) = f ∫ Ca(τ) RF(t- τ) dτ
tissue
Deconvolution :
find f and RF(t)
f
time
HBWL
Conclusion
Measure the
tissue conc
Measure the input conc i.e.
input function
Bolus input : Ctis(t) = f Ca(0) Δt RF(t)
Estimate f and RF(t)
Veneous injection : Ctis(t) = f ∫ Ca(τ) RF(t- τ) dτ
HBWL
Deconvolution ~ Modelbased
•  Use a model e.g.: Monoexponentiel, biexponentiel,
•  Optimise the free parameters by least square fit to
tissue enhancement curve
•  It is robust
•  Relative insensitive to noise
•  Incorrect if the model is inappropriately chosen
HBWL
Deconvolution ~ Modelfree
• 
• 
• 
• 
• 
• 
No model a priory
Very flexible: many of free parameters
A projection
Very sensitive to noise
Incorrect if not regularized rigoriously
Fourier transform, SVD, GSVD, Tikhonov, GPD
HBWL
The story can begin
HBWL
DCE T1 perfusion and blood brain barrier:
Methodological considerations and applications
January 24. 2012
by
Henrik BW Larsson
Functional Imaging Unit, Diagnostic Department
HBWL
Acute MS lesions
T1
T1-Gd
HBWL
Paramagnetic Contrast, Gd-DTPA
• Decreases homogeneity
in molecular environment
• Increases relaxation
speed!
HBWL
Acute MS lesions
T1
T1-Gd
HBWL
BBB permeability : PS / Ktrans / Ki
HBWL
Dynamic Contrast Enhanced
Heart Perfusion, normal
MR signal
blood
tissue
time
HBWL
Dynamic Contrast Enhanced
Heart Perfusion, normal
MR signal
blood
tissue
time
HBWL
Rest scan : healthy subject
HBWL
Stress scan: healthy subject
HBWL
Gadolinium-DTPA
as perfusion CA
T1w MR-signal
T2*w MR-signal
extravascular
intravascular
Time to peak
time
bolus
time
bolus
HBWL
Gadolinium-DTPA
as perfusion CA
T1w MR-signal
T2*w MR-signal
extravascular
intravascular
Time to peak
time
bolus
time
bolus
HBWL
Dynamic T1 weighted contrast enhanced
perfusion MRI in brain at 1.5 tesla
F = 83 ml/100/min
Larsson at al MRM 2001, p272
HBWL
Dynamic T1 weighted contrast
enhanced perfusion MRI in brain at
3 tesla
Time resolution:
5 slices pr. 1.24s
HBWL
Dynamic T1 weighted contrast
enhanced perfusion MRI in brain at
3 tesla
Time resolution:
5 slices pr. 1.24s
HBWL
Gd gives shorter T1 = faster recovery
•  Short range – works only in
same compartment
•  Increased signal in blood and
tissue on T1W images
•  Normally BBB - Ktrans
•  Now perfusion - CBF
HBWL
MRI Perfusion Measurement
Procedure, Gd:
• 
• 
• 
• 
Intravenous injection of Gd-DTPA
Rapid imaging using T1W FFE (SR TurboFLASH)
Conversion of signal units to concentration
via relaxation rate units
Data analysis: deconvolution approach
HBWL
From MR signal to concentration of
contrast agent
R1 = 1/T1
Concentration
MR signal
Ctis Ca
callibration by
external phantoms
HBWL
R1Gd+ - R1Gd- = relaxivity • C
ΔR1 (t) = relaxivity • C(t)
HBWL
Conversion of MR signal to tracer concentration
Measured
MR signal
Saturation
recovery delay
Equilibrium
magnetization
Tissue R1
before bolus
Flip angle
Change of tissue R1
due to Gd bolus
s(t) = M0sin(α)[1-exp(-TD(R1+ΔR1(t)))] , ΔR1(t) = r1c(t)
Saturation recovery equation
M0sin(α)
Relaxivity of
contrast agent
Tracer
concentration
s(TD)
1/R1
HBWL
TD
T1 measurement
s(t) = M0sin(α)[1-exp(-TD· R1)]
T1 = 1.23 s
(frontal
gray matter)
R1
M0
HBWL
Favorite brain arteries for AIF
Middle Cerebral Artery (MCA)
ICA
Internal Carotid Artery (ICA)
HBWL
Slice position in DCE
HBWL
Dynamic T1 weighted contrast
enhanced perfusion MRI in brain at
3 tesla
Sat-Recovey:
TI = 120ms
TR= 4 ms
TE= 2 ms
Angle = 30o
Voxel = 3x3x6 mm
Time resolution:
5 slices pr. 1.24s
Dose:
Magnevist/
Dotarem
0.05mmol/kg
HBWL
Dynamic T1 weighted contrast
enhanced perfusion MRI in brain at
3 tesla
Sat-Recovey:
TI = 120ms
TR= 4 ms
TE= 2 ms
Angle = 30o
Voxel = 3x3x6 mm
Time resolution:
5 slices pr. 1.24s
Dose:
Magnevist/
Dotarem
0.05mmol/kg
HBWL
Deconvolution approach to quantitative
perfusion measurement
arterial input
tissue response
⊗
t (s)
RF(t)
=F
ca(t) (mM)
ct(t) (mM)
tissue curve
t (s)
t (s)
t
Ct(t) = F· Ca(t) ⊗ RF(t) = F · ∫ Ca(t’) RF(t-t’) dt’
0
MTT = ∫ RF(t)dt
CBV = F· MTT
HBWL
Dynamic T1w contrast enhanced MRIperfusion
CBF map
HBWL
Deconvolution methods
gray matter ROI
arterial input
ct(t) = F· ca(t) ⊗ RF(t)
RF(t):
monoexp
tissue curve
RF(t): deconvolution
with SVD
RF(t):
deconvolution with
Tikhonov
HBWL
The residue impulse response function
gray matter ROI
arterial input
tissue curve
RF(t): deconvolution
with SVD
RF(t): deconvolution
with Tikhonov
HBWL
Tikhonov’s deconvolution
or
Generalized SVD
large
λ
small
Larsson at al JMRI 2008, p754
HBWL
Examples of CBF maps
ml/100g/min
120
100
80
60
40
20
We found perfusion value for ROI’s to be
62 ml/100g/min in gray matter and
21 ml/100g/min in white matter
in 7 patients with acute optic neuritis.
HBWL
Combined Anatomical and Functional
MRI: Anatomy & perfusion of the brain
ml/100g/min
HBWL
Two patients with stroke (one week old)
T2- w image
DWI
Larsson at al JMRI 2008, p754
CBF
ml/100g/min
HBWL
CBV
ml/100g
MTT
s
10
40
8
32
6
24
4
16
2
8
0
0
10
40
8
32
6
24
4
16
2
8
0
Larsson at al JMRI 2008, p754
0 HBWL
Perfusion in clinical desicision
making
•  Assesment of cerebral ischemia
•  Assesment of reversible versus
irreversible tissue damage before
treatement
HBWL
Comparison of MR perfusion
methods:
Carotic flow measurement (PCM)
DCE
ASL
PET (15H2O)
Otto Henriksen at al JMRI 2012, in press
HBWL
HBWL
17 healthy subjects, 20-30 years
Performed twice - Variation: Within persons – method variation
Between persons variation
HBWL
Table 1. Cerebral blood flow measurements by different modalities
Bias *
Mean
Global CBF
Gray matter CBF
White matter
CBF
pvalue
sbetw
swith
CVbetw
CVwith
**
PCM
65.19†
0.43
0.813
11.32
4.80
17.4%
7.4%
ASL
37.01
-0.30
0.762
6.00
1.76
16.2%
4.8%
DCE
42.99
-3.77
0.100
8.60
6.47
20.0%
15.1%
PET
41.88
0.69
0.757
6.69
4.99
16.5%
11.9%
ASL
50.15
-0.00
0.998
9.42
3.05
18.8%
6.1%
DCE
65.55‡
-5.90
0.095
12.83
10.01
19.6%
15.3%
PET
58.55
0.55
0.859
10.12
6.91
17.3%
11.8%
ASL
21.72
-0.40
0.290
3.53
1.07
16.3%
4.9%
DCE
16.75‡
-1.23
0.186
5.61
2.64
33.5%
15.8%
PET
22.57
0.87
0.487
3.41
2.80
15.1%
12.4%
HBWL
Table 1. Cerebral blood flow measurements by different modalities
Bias *
Mean
Global CBF
Gray matter
CBF
White matter CBF
pvalue
sbetw
swith
CVbetw
CVwith
**
PCM
65.19†
0.43
0.813
11.32
4.80
17.4%
7.4%
ASL
37.01
-0.30
0.762
6.00
1.76
16.2%
4.8%
DCE
42.99
-3.77
0.100
8.60
6.47
20.0%
15.1%
PET
41.88
0.69
0.757
6.69
4.99
16.5%
11.9%
ASL
50.15
-0.00 0.998 9.42
3.05
18.8%
6.1%
DCE
PET
65.55‡
58.55
-5.90 0.095 12.83
0.55 0.859 10.12
10.01
6.91
19.6%
17.3%
15.3%
11.8%
ASL
21.72
-0.40
0.290
3.53
1.07
16.3%
4.9%
DCE
16.75‡
-1.23
0.186
5.61
2.64
33.5%
15.8%
PET
22.57
0.87
0.487
3.41
2.80
15.1%
12.4%
HBWL
Comparison of MR perfusion
methods:
Otto Henriksen at al JMRI 2012, in press
HBWL
Conclusion
• 
• 
• 
• 
• 
• 
Large inter-individual variability – all methods
Same mean global CBF across ASL, DCE, PET
PCM global flow > global ASL, DCE, PET
No correlation between methods, except
A (weak) correlation between PCM and DCE
Existence of significant subject-method interaction
HBWL
Does it works if BBB is leaky?
• 
• 
• 
• 
Can we still estimate CBF?
Can we estimate PS (Ki or Ktrans)?
Can we estimate CBV ?
Can we differentiate between Vd and CBV?
HBWL
Blood brain barrier defect in brain tumors
After CA
Before CA
Tikonov: CBF = 53ml/100g/min
90
Gray matter
80
Tikonov: CBF = 157 ml/100g/min
350
Tumor lesion
300
70
250
Concentration a.u.
Concentration a.u.
60
50
40
30
20
200
150
100
50
10
0
0
-10
0
20
40
60
80
100
time (sec)
120
140
160
180
-50
0
20
40
60
80
time (sec)
100
120
160
HBWL
140
180
Blood brain barrier defect in brain tumors
After CA
Before CA
Tikonov: CBF = 53ml/100g/min
90
Gray matter
80
Tikonov: CBF = 157 ml/100g/min
350
Tumor lesion
300
70
250
Concentration a.u.
Concentration a.u.
60
50
40
30
20
200
150
100
50
10
0
0
-10
0
20
40
60
80
100
time (sec)
120
140
160
180
-50
0
20
40
60
80
time (sec)
100
120
160
HBWL
140
180
Permeability - surface area map
ml/100g/min
50
25
0
HBWL
Measurement of Brain Perfusion, Blood
Brain Barrier Permeability using
Dynamic Contrast Enhanced T1Weighted MRI at 3 T
Larsson at al MRM 2009, p1270
HBWL
Compartment model
BBB
arteries
Ca(t)
veins
F
F
blood
Cb(t)
Vb
Ki
Ki(1-Hct)
tissue
Ce(t)
Ve
Intracellular
space
Vtis
Vd
HBWL
BBB
arteries
Ca(t)
veins
F
F
blood
Cb(t)
Vb
Ki
Ki(1-Hct)
tissue
Ce(t)
Ve
Intracellula
r
space
Vtis, Ctis
Vd
Tissue
voxel
HBWL
HBWL
HBWL
HBWL
Patlak & Gjedde method
HBWL
Calculation of BBB permeability
Patlak & Gjedde method
or
two-compartment model & Tikhonov
CBF estimation
HBWL
HBWL
HBWL
HBWL
Gray matter
Tumor
Tumor
HBWL
Does it works if BBB is leaky?
• 
• 
• 
• 
Can we still estimate CBF?
Can we estimate PS (Ki or Ktrans)?
Can we estimate CBV ?
Can we differentiate between Vd and CBV?
yes
yes
yes
yes
In the relevant range !
HBWL
F (ml/100g/min)
120
Ki (ml/100g/min)
5
100
4
80
3
60
3
40
2
20
1
0
0
Anatomy, T2w
Vd (ml/100g)
20
CBVVb (ml/100g)
20
17
17
13
13
10
10
7
7
3
3
0
0
HBWL
F (ml/100g/min)
100
Ki (ml/100g/min)
5
83
4
67
3
50
3
33
2
17
1
0
0
Anatomy,
T2w
T2 w TSE
Vd (ml/100g)
20
CBVVb (ml/100g)
20
17
17
13
13
10
10
7
7
3
3
0
HBWL
0
DCE in evaluation of tumor
recurrence versus radiation
necrosis
FDG-PET as referene
Vibeke A Larsen et al Submitted
HBWL
Tumor
recurrence
Tumor
recurrence
Radiation
necrosis
HBWL
Recurence versus necrosis
HBWL
Recurence versus necrosis
HBWL
DCE in evaluation of tumor
recurrence versus radiation
necrosis
FDG-PET as referene:
CBV from DCE seems a sensible parameter
Larsen VA at al Submitted
HBWL
Pivotal for all measurement
The input function
Partial volume effect on the arterial input
function in T1-weighted perfusion imaging and
limitations of the multiplicative rescaling
approach
Adam E Hansen et al. MRM 2009, p1055
HBWL
Effect of partial volume on the arterial input function
high in-plane
resolution:
(1.15mm)2
voxel
HBWL
Point spread function of input function
HBWL
HBWL
Future directions
•  The feasibility at 7 T ?
•  Incorporation of water exchange,
i.e. water permeability
HBWL
Larsson at al MRM 2001, p272
T1 versus T2 Gd based perfusion
Advantage T1
• 
• 
• 
• 
No image distortion (no susceptibility)
Input function clearly defined
No bias due to defect BBB
Half of normal clinical dose
Disadvantage T1
• 
• 
• 
• 
• 
Fewer slices
Lower S/N for tissue conc time function
Necessitate high field strength 3 tesla
Injection of contrast agent
Can only be repeated a few times
Advantage T2
•  Many slices
•  High S/N for the tissue conc time function
Disadvantage T2
• 
• 
• 
• 
• 
• 
Input function cannot be defined clearly
Perfusion is estimated to high
MR signal to conc is problematic
Perfusion is bias when BBB is defect
Injection of contrast agent
Repeatable ?
HBWL
Conclusion
•  It is possible to generate CBF maps using DCE T1 weighted MRI at 3 T
•  Perfusion values are consistent with literature
•  Tikhonov’s method is best suited for deconvolution
•  DCE T1 weighted MRI appears promising for distinguishing tumor
recurrence and radiation necrosis employing CBV
•  Easy identification of vasculature with DCE T1 weighted MRI allows to
study details of input function
HBWL
Thanks to
Egill Rostrup, Adam E Hansen, Otto Henriksen, Glostrup Hospital
Bente Sonne Møller, Helle Juhl Simonsen, Marjut Lindhart, Glostrup Hospital
Olav Haraldseth, Trondheim
Vibeke Andrée Larsen, Ian Law, Julie M Grüner, RH
Frederic Courivaud, Philips Clinical Scientist
Lundbeck Centre for Neurovascular Signaling
Thank you for your attention !
HBWL

Similar documents