Integrated PET/MRI
Ciprian Catana, MD, PhD
ccatana@nmr.mgh.harvard.edu
1
Outline
PET/MRI:
Brief history and current state-of-the-art
Methodological advances enabling new
science:
What MR can do for PET
What PET can do for MR
Potential research and clinical
applications
“Physics” motivation for simultaneous PET/MRI positron range error reduction in magnetic fields
e+
ERROR
DUE TO
POSITRON
RANGE
511 KEV
PHOTON
e– e+
ANNIHILATION
Representative paths in tissue of a 2 MeV electron
511 KEV (Raylman et al. Int J Rad Onc Biol Phys, 1997; 37(5))
PHOTON
*In-plane
PET
scanner
spatial
resolution might improve at high
magnetic fields (i.e. B>7T), for high
energy positron emitters (i.e. 82Rb,
Et=3.15 MeV) if it is not dominated by
other factors (i.e. crystal size, noncolinearity, etc.)



F Lor = q V x B
0.334
R=
(2m p Et ) + Et2
B
q – particle charge; V – positron velocity; B –
magnetic field; Et – component of the positron
kinetic energy (MeV) perpendicular to the
magnetic field;
mp – rest mass of the positron
Integrating PET and MRI does not mean placing
an existing PET scanner inside an MR scanner.
PET effects on the MR:
No ferromagnetic components allowed;
Disturb homogeneity of the B0 field;
RF interference with the MR Tx/Rx coils;
Susceptibility artifacts and eddy currents.
…
MR effects on the PET:
PMTs very sensitive to magnetic fields;
RFI, heating, vibrations, etc.
…
General considerations:
Space constraints inside the MR;
Cost !
Integrated PET/MR scanners have been
developed for small animal and human imaging.
2011
(Wehrl HF, Judenhofer MS, Wiehr S, Pichler BJ,
Eur J Nucl Med Mol Imaging 36 (Suppl 1): S56-58; 2009, updated)
PET-MR(-CT) Scanners Available for Human Use
Philips Ingenuity TF PET/MR
GE PET/CT
Siemens Biograph mMR
Patient Transport
GE MR
Outline
Integrated PET/MRI:
Brief history and current state-of-the-art
Methodological advances enabling new
science:
What MR can do for PET
What PET can do for MR
Potential research and clinical
applications
What MR can do for PET
3. Partial volume effects corr.
2. Motion correction
4. Arterial input function estimation
6.E-03
5.E-03
Counts (a.u.)
1. Attenuation correction
4.E-03
3.E-03
2.E-03
1.E-03
0.E+00
0
200
400
600
Annihilation photons can interact with the
subject before reaching the detectors.
I ( x) = I (0)e x
p2 = e  ( D x )
2
μ
pcoinc. = p1 p2 = e x e  ( D x ) = e D
1
D
p1 = e  x
xx
Uniform cylinder before and after AC (Ø=20
cm, µ=0.096 cm-1)
Atten . Corr. = e
D
Estimating the tissue linear attenuation
correction factors from MR is difficult.
A number of factors have to be considered for implementing an accurate
MR-based attenuation correction method.
Separating the bone from air-filled cavities is the most challenging task using
conventional MR sequences.
MR
Scaled CT
Segmented CT
10
Ultra-short echo time (UTE) sequences
can be used for bone imaging.
TE = 0.07 ms
TE = 2.46 ms
11
MR-based PET attenuation correction
using DUTE data
ATTENUATION MAPS
CTsegmented (top) and DUTEsegmented
(bottom)
RECONSTRUCTED PET IMAGES
Attenuation correction factors derived from the
CTsegmented (top) and DUTEsegmented (bottom).
(C. Catana et al, J Nucl Med, 2010)
12
Skull/soft tissue/air segmentation can be
achieved from DUTE and MPRAGE data.
Segmented CT
Atlas-T1-DUTE
Clare Poynton
13
Atlas-based methods allow generation of
continuous-valued attenuation maps.
MPRAGE
David Izquierdo
14
The attenuation caused by the RF coils has to
be accounted for in an integrated scanner.
50
-50
BrainPET
gantry
Coil attenuation
ignored
RF coil
5
Patient
table
-5
3 mm trans.
along X axis
3 mm trans.
along Y axis
no coil
3 mm Y
BrainPET prototype inside the 3T MR
scanner (MGH installation)
3°rot.
about Z axis
3 mm X
3 deg Z
1.E+06
1.E+05
1.E+04
1.E+03
1.E+02
1.E+01
-60
-50
-40 -30 -20 -10
0
Relative change (%)
10
20
Relative change images and histograms
15
Motion is difficult to avoid in long PET
studies.
image blurring/artifacts;
attenuation/emission data mismatch;
inaccurate quantification.
Spatial resolution “loss”
16
Various motion correction approaches have
been investigated for neuroPET studies.
3. Event-by-event correction
1. Eliminate head motion
2. Inter-frame correction
17
Navigators introduced in standard MR sequences
provide high temporal resolution motion estimates.
( Andre J.W. van der Kouwe et al, Magnetic
Resonance in Medicine 2006; 56: 1019-1032 )
MR before and after MC
18
Motion correction algorithm for dynamic studies
on the BrainPET prototype.
(C. Catana et al, JNM, 2011)
19
MR-derived motion estimates can be used
to retrospectively correct the PET data.
PET data before (left) and after (right) MR-assisted motion correction
•
•
•
•
Healthy volunteer;
~5 mCi 18FDG;
Simultaneous MR-PET study;
PET MC applied in LOR space.
Motion estimates derived from EPI
MR series every 3 seconds
(C. Catana et al, JNM, 2011)
20
Methods: Phantom Study in PET-MR
Phantom
PET
External
trigger
In and out pressure
from ventilator
MR
Vent
“Beating Phantom”
Gel with
background activity
Inflatable balloon
Hot spheres attached to balloon (10mm diameter)
Guerin and El Fakhri, Med Phys 2011
Methods: Tagged MR
GRE sequence
GRE sequence with tagging
Tagging patterns provide additional motion information.
Courtesy G. El Fakhri, MIPI, MGH
Phantom Results: Reconstruction
Uncorrected
Gated
90
C=1.2
σ=19%
C=1.7
Sphere 2:
Diameter 10mm
Resolution 2mm each
Max motion ~3.0mm
(small motion)
σ=48%
MR motion corrected
C=2.0
σ=20%
Chun, El Fakhri, JNM 2011
0
Primate Results
• Motion Correction with Primate in simultaneous PET-MR
Gated tagged MR
Chun, El Fakhri, JNM 2011
Gated PET
Primate Results
Chun et al. JNM 2012, 53(8):1284-1291
Partial volume effects (PVE) affect PET
data quantification.
original
3 mm
5 mm
7 mm
100
90
80
70
60
50
40
30
20
10
0
To account for PVEs, information about the size of the structures of interest and
the spatially variant point spread function of the PET scanner is needed.
26
The high resolution morphological MRI
data can be used for PET PVE correction.
Automated brain structures segmentation from the MPRAGE data.
27
The high resolution morphological MRI
data can be used for PET PVE correction.
Morphological MR
(ME-MPRAGE)
Recovery
coefficients
Original PET
PET after
regional
PVE correction
Spencer Bowen, MGH 28
6.E-03
Counts (a.u.)
5.E-03
The non-invasive estimation of the radiotracer
input function can be improved using MR data.
20
4.E-03
3.E-03
2.E-03
1.E-03
40
0.E+00
0
200
400
600
50
Time (sec)
60
80
100
100
120
150
140
TOF MR
200
50
100
150
ME-MPRAGE
200
250
50
100
150
200
250
Dynamic PET frames
FDG AIF
Daniel Chonde, MGH
29
What MR can do for PET
MR-based
attenuation corr.
MR-based
motion corr.
Accurate tissue time activity curves
&
radiotracer AIF estimate
Image-based AIF
estimation
Partial volume
effects corr.
Compartment model
&
Kinetic analysis
Improved PET data quantification
30
What PET can do for MR
PET-guided MR imaging
In-vivo quantification of “smart” MR probes
Techniques cross-calibration and validation
31
Fibrin Targeted Gd-based Contrast Agent
Fibrinspecific
Peptide
Bis Gd
chelate
Bis Gd
chelate
Linker
O
O
O OH2
O
O
O OH2
O
N
N
N Gd
O
N
O
O
O
O
OH
O
O
H
N
N
H
N
H
O
HN
O
O
O
N
O
O
N
Gd
N
N
N
H
O
H
N
O
O
N
H
H
N
O
O
N
H
O
H
N
O
OH
O
N
H
O
O
NH
OH
O
O
O
O
N
Gd
N
N
N
O
OH2
EP-2104R
O
NH
HN
N H2
O
O
O
O
O
O
N
O
H
N
Cl
O
O
S
OH
HN
Na 5
N
N Gd
O
N
O
O
S
N
O
O
OH2
O
O
Improved fibrin affinity. High relaxivity: 4 Gd + protein binding.
No metabolism issues. Minimal Gd retention.
Overoye-Chan et al., J Am Chem Soc 2008, 130:6025-39
32
Embolic Stroke Model
A
B
Image pre contrast
agent
Image post contrast
agent
C
Arterial anatomy
(Ritika Uppal, Ilknur Ay, Peter Caravan)
33
High sensitivity PET can guide the high
resolution MR study.
Cu
Gd
Peptide
Gd
Gd
Fibrin-targeted MR-PET Probe
Whole-body PET “screening”
Uppal, Caravan, Radiology, 2011;258(3):812-20
Focused MR study
34
Relaxivity is sensitive to many molecular factors
that can be modulated to induce an MR signal.
Enzyme activation  change property of contrast agent
A) GFP mRNA injection in R; B) EgadMe, β-gal injection in
R; C) β-gal staining
(Louie, Meade et al, Nature Biotech, 2000)
35
pH-responsive “smart” MR probe
Low pH: more exchangeable protons (blue)
proton exchange rate ideal
High pH: fewer exchangeable protons (blue)
proton exchange too fast
(Kalman, Sherry, Caravan et al, Inorg Chem 2007, 46,5260;
Ali, Sherry, Caravan et al, Chem Eur J 2008, 14:7250-8)
36
The MR signal depends on both relaxivity and
contrast agent concentration.
Two unknowns, one measurement
1
1
= r1 [Gd ] + 0
T1
T1
relaxivity
contrast agent
concentration
37
A “smart” dual-modal MR-PET agent can
quantitatively and non-invasively measure pH.
Dual
probe
8
pH (MR-PET)
Responsive Probe
MR
y = 1.01x
R2 = 0.95
7.5
7
6.5
PET
6
6
6.5
7
7.5
8
pH (electrode)
Quantification
(L. Frullano, C. Catana, T. Benner, A. Sherry, P.
Caravan. Angew. Chem. Int. Ed. 2010; 49:2382-4)
38
Cross-validate PET and MRI cerebral perfusion
measurements.
Arterial spin labeling technique
Bolus tracking technique
(L Ostergaard. JMRI 2005, 22:710-7)
[O15]H2O PET and steady-state ASL MR
(F.Q.Ye et al. MRM 2000; 44:450-456)
(Y Ozsunar and AG Sorensen et al;
Topics in MRI 2000; 11(5): 259-272)
39
Validate and model the relationship between
OEF and BOLD signal during neuronal
activation.
(H. Ito et al. J Cerebral Blood Flow and Metab. 2005; 25: 371-377)
40
MR and PET Can Help Each Other
MR-based
attenuation corr.
MR-based
motion corr.
Image-based AIF
estimation
Partial volume
effects corr.
Improved PET data quantification
PET-guided
MR imaging
Quantification of
MR probes
Techniques
cross-calibration
Multimodal quantitative MR-PET imaging
41
Automated Data Processing and Analysis
Dan Chonde, MGH
42
Outline
Integrated PET/MRI:
Brief history and current state-of-the-art
Methodological advances enabling new
science:
What MR can do for PET
What PET can do for MR
Potential research and clinical
applications
43
Potential Applications
How can MR-PET …
1) increase
increase diagnostic
diagnosticaccuracy?
accuracy?
oncology, cardiology, neurology
2) improve patient experience?
3) advance scientific discovery?
“Indications in which PET/MRI may be favorable
over PET/CT, depending on tumor entity”
PET/MRI is relevant for (favorable to PET/CT) …
Most frequent site of
metastases*
Tumor entity
Brain
Lung
Liver
Bone
Staging
category
Head and neck SCC
−
+
−
+
T
Extracapsular spread; bone infiltration
Non-small cell lung
cancer
+
+
+
+
M
Distant metastases
Breast cancer
+
+
+
+
T/M
HCC
−
+
+
+
T
Colorectal
carcinoma
−
+
+
−
T/M
Circumferential resection margin; liver metastases; tumor
regression rate to neoadjuvant therapy
Soft-tissue sarcoma
−
+
+
−
T/M
Tumor size and depth of infiltration defines T category;
muscular, neurovascular, and bone invasion
Primary bone
tumors
−
+
−
−
T
Presurgical evaluation (e.g., neurovascular invasion);
exact tumor size and response to neoadjuvant treatment
Melanoma
+
+
+
+
M
Exact number and location of metastases for presurgical
evaluation
M
Extranodal dissemination; early therapy response
assessment
Lymphoma
Special objective/prognostic factor
Primary diagnosis and T-stage (benefit compared with
PET/CT; potential benefit over MRI mammography alone
is questionable); distant metastases
Pretransplantation evaluation
*Frequency of metastatic spread (frequently [+], rare [−]) is according to AJCC Cancer Staging Manual, seventh edition; PET/CT and
PET/MRI are considered equally accurate for N-staging, and thus importance of N-staging is not discussed.
C. Buchbender et al. Oncologic PET/MRI, Part 1 & 2. JNM 2012; 53: 928-938 &1244-1252
Head & Neck MR-PET
More precise metabolic-anatomic allocation of the FDG-avid lesion
FDG-PET/CT
54-y-old man with
gingival SCC arising
from maxilla
FDG-PET/MR
C. Buchbender et al.
J Nucl Med 2012; 53:928
938
Breast MR-PET
Accurate spatial registration
MRscan 1
PETscan 1
MRscan 2
Spencer Bowen, MGH
Fused MRscan 1-PETscan 1
Fused MRscan 2-PETscan 1
Liver PET/CT + MRI
Increased lesion conspicuity
P. Veit-Haibach, F.P. Kuhn, F. Wiesinger, G. Dalso and G. van Schulthess.
Magn Reson Mater Phy (2013) 26: 25-35
Pelvis MR-PET
Improved soft tissue discrimination and functional information
T1-weighted post-contrast MRI
DWI MRI
PET/MRI fusion
FDG PET
ADC map
Collaboration with A. Guimaraes, MGH
Whole-body PET/MRI
Total body bone scan with 219 MBq 18F-NaF of a
61-year-old female (76 kg), with 80 s per bed
position for PET acquisition and 105 min uptake
time.
Images courtesy of HZDR, Dresden
Data acquired using the Philips WB PET/MR
scanner
Magn Reson Mater Phy (2013) 26:5-23
“Advantageous Features of Combined PET/MR
with Focus on Cardiac Applications”
C. Rischpler, S.G. Nekolla, I. Dregely and M. Schwaiger. Hybrid PET/MR Imaging of the
Heart: Potential, Initial Experiences, and Future Prospects. JNM 2013; 54:402-415
Myocardial Perfusion MR-PET Study
13NH
3
PET
Fused
MR-PET
Gd-DTPA
MR
C. Rischpler, S.G. Nekolla, I. Dregely and M. Schwaiger. Hybrid PET/MR Imaging of the
Heart: Potential, Initial Experiences, and Future Prospects. JNM 2013; 54:402-415
52
Simultaneous 13NH3-PET/MR study
demonstrates stress-induced ischemia.
13NH -PET
3
Stress
LGE MRI
Fused LGE MRI with
13N-PET Stress
Anterior/anterolateral ischemia (PET) without significant delayed contrast enhancement (MRI).
Left anterior descending artery disease was subsequently confirmed by catheterization.
Courtesy of J. Lau, R. Laforest, P. Woodard
53
Simultaneous FDG-PET and DCE MR study
allows myocardial viability assessment.
MR
FDG-PET
Fused PET/MR Cine
Delayed contrast enhancement MR and FDG-PET images acquired in diastole.
Fused cine created from the PET list mode data binned into 8 phases fused with
simultaneously acquired free-breathing real time SSFP cardiac cine.
[J. Lau, R. Laforest, S. Sharma, J. McConathy, A. Priatna,
L. Amado, R. Gropler, P. Woodard. ISMRM 2013, Oral: #0573]
54
Potential Benefits
How can MR-PET …
1) increase diagnostic accuracy?
2) improve patient experience?
3) advance scientific discovery?
MR-PET can improve patient experience
• Two exams in one session:
– Increased patient compliance
– Reduced need for sedation/anesthesia in pediatric
patients
– One pharmacological challenge for two exams
• Reduced radiation exposure:
– Pediatric patients, women of childbearing age,
chronic patients
The radiation dose from CT vs FDG-PET is
significantly higher in children compared to adults.
(F.H. Fahey, “Dosimetry of Pediatric PET/CT”. J Nucl Med
2009; 50: 1483-1491)
(H. Jadvar et al, “PET and PET/CT in Pediatric
Oncology”. Semin Nucl Med 2007; 37:316-331)
F.W. Hirsch, B. Sattler et al. University of Leipzig
15-yo boy with left testicular tumor with retroperitoneal,
supraclavicular, hepatic and lung metastases
Potential Benefits
How can MR-PET …
1) increase diagnostic accuracy?
2) improve patient experience?
3) advance scientific discovery?
Simultaneous MR-PET opens new opportunities
for studying the brain.
(Figure from “Imaging of the Human Brain in Health and Disease”, J.E. Johnson ed.)
61
Neuropsychiatric conditions contribute the most to the overall
burden of non-communicable disease, more than either
cardiovascular disease or cancer.
Contribution by different non-communicable diseases to DALYs worldwide in 2005
DALY – disability-adjusted life-year (sum of the years lived with disability and years of life lost)
Prince M et al. No health without mental health. Lancet 2007; 370(9590): 859-77
PET and MRI provide complementary information about
the brain.
PET RADIOTRACERS
MR TECHNIQUES
Hemodynamic parameters:
cerebral blood flow (H215O, 15Obutanol, 13NH3,…), cerebral blood
volume (11CO)
Substrate metabolism:
glucose (18F-FDG), oxygen (15O2)
Protein synthesis:
11C-methionine, 11C-leucine, 11Ctyrosine
Amino acid transport:
18F-fluoroethyltyrosine, 18Ffluorophenylalanine,…
Nucleosides and DNA synthesis:
18F-fluorothymidine,…
Neurotransmitter biochemistry:
precursors (18F-FDOPA, 11C-AMT,…),
transporters (11C-methylphenidate,
11C-cocaine,…), receptors (11Craclopride, 11C-nicotine, 18Faltanserin,…), enzyme activity (11Cdeprenyl, 11C-donepezil,…)
…
Anatomy:
high resolution morphology,
angiography
Perfusion:
cerebral blood flow and blood volume,
mean transit time, time to peak,
relative vessel size and permeability,…
Water diffusion:
mean diffusivity, fractional anisotropy,
apparent diffusion coefficient, fiber
orientation
Brain function:
BOLD contrast, PWI
Chemical composition:
1H-MRS (NAA, Cr, Cho, Lac, mI) , 31PMRS
…
MR-PET allows the assessment of
anatomical/functional/molecular changes in
dementia.
Surface projections
show areas with
reduced metabolism
FDG
PET
Fused
MR-PET
DTI
MPRAGE
Collaboration with
Brad Dickerson,
Alexander Drzezga
C. Catana et al. “PET and MR Imaging: The odd couple or a match
made in heaven?” JNM 2013; 54:1-10
65
MR-PET could help us understand the mechanism
of action of therapeutic agents in GBM patients.
MRI
PET
Fused
MRI-PET
[18F]FDG
[18F]FMISO
[18F]FLT
[18F]FET
(C. Catana et al. “PET/MRI for Neurological Applications”. JNM 2012; 53:1916-1925 )
66
Cross-calibration measures could elucidate the
mismatch-penumbra debate in ischemic stroke patients.
Sobesky et al Stroke 2005; 36: 980-985
67
fMRI-PET Assessment of the Response to a
(a)
Physiological
Challenge (i.e. experimental
pain)
Brain regions
showThalamus
fMRI-PET activation overlapx = -4
x = -14
(a)
Putamen/NAc
y = 14
r = 0.96
p = 0.001
0.0
-0.1
0.0
0.2
0.1
0.0
r = -0.28
p = 0.5
0.2
0.3
ΔBPND
0.4
-0.4
-0.2
Putamen/NAc (R)
0.3
Z
2.3
Putamen/NAc
-0.1
0.1
PET
0.1
3
fMRI
0.2
0.3
y=4
0.0
0.2
0.4
0.6
ΔBPND
3
0.2
Wey, Gollub, Kong et al.
ET
0.2
ΔBOLD (% Signal Changes)
0.3
Thalamus (L)
0.3
x = -4
MRI
ignal Changes)
(b)
Putamen/NAc (R)
Thalamus (L)
ignal Changes)
y = 14
(b) Thalamus
ΔBOLD (% Signal Changes)
x = -14
y=4
Z
68
FDG-PET/MRI Study of Dexmedetomidine
Sedation
pASL-MRI
FDG-PET
Baseline
Baseline
After sedation
After sedation
Seun Akeju, MGH
70
Acknowledgements
Massachusetts General Hospital
Bruce Rosen
Daniel Chonde, Spencer Bowen, Kevin Chen, David Izquierdo
Jacob Hooker, Grae Arabasz, Shirley Hsu, Steve Carlin, Chris
Moseley
Alexander Guimaraes, Andre van der Kouwe, Elizabeth
Gerstner, Tracy Batchelor, Peter Caravan, Larry Wald, Brad
Dickerson, Alexander Drzezga, Jian Kong, Randy Gollub,
Darin Dougherty, Monica Wey, Joe Mandeville, Christin
Sander, …
Siemens Healthcare
Greg Sorensen
Larry Byars, Christian Michel, Matthias Schmand, …
Keith Heberlein, Michael Hamm, Thomas Benner, Matthias
Fenchel, …
University of California Davis
Simon Cherry (PhD advisor)
71