Peter A. Bandettini

Unit on Functional Imaging MethodsLaboratory of Brain and Cognition

&Functional MRI Facility

Neuronal Current Imaging(GRC: In vivo Magnetic Resonance 2004)

Natalia Petridou

Jerzy Bodurka

Peter A. Bandettini

Unit on Functional Imaging Methods /Functional MRI Facility, NIMH, NIH

Afonso C. SilvaLaboratory of Functional and

Molecular Imaging, NINDS, NIH

Dietmar PlenzUnit of Neural Network Physiology,

NIMH, NIH

Primary researchers and collaborators:

Related Papers

J. Bodurka, P. A. Bandettini. Toward direct mapping of neuronal activity: MRI detection of ultra weak transient magnetic field changes. Magn. Reson. Med. 47: 1052-1058, (2002).

M. Singh, Sensitivity of MR phase shift to detect evoked neuromagnetic fields inside the head.IEEE Transactions on Nuclear Science. 41: 349-351, (1994).

D. Konn, P. Gowland, R. Bowtell, MRI detection of weak magnetic fields due to an extended current dipole in a conducting sphere: a model for direct detection of neuronal currents in the brain. Magn. Reson. Med. 50: 40-49, (2003).

J. Xiong, P. T. Fox, J.-H. Gao, Direct MRI Mapping of neuronal activity. Human Brain Mapping, 20: 41-49, (2003)

G. C. Scott, M. L. Joy, R. L. Armstrong, R. M. Henkelman, RF current density imaging homogeneous media. Magn. Reson. Med. 28: 186-201, (1992).

H. Kamei, J, Iramina, K. Yoshikawa, S. Ueno, Neuronal current distribution imaging using MR. IEEE Trans. On Magnetics, 35: 4109-4111, (1999)

M. L. Joy, G. C. Scott, R. M, Henkelman, In vivo detection of applied electric currents by magnetic resonance imaging. Magn Reson Imaging 7: 89-94, (1989).

J. Bodurka, P. A. Bandettini. Current induced magnetic resonance phase imaging. Journal of Magnetic Resonance, 137: 265-271, (1999).

R. Chu, J. A. de Zwart, P. van Gelderen, M. Fukunaga, P. Kellman, T. Hollroyd, J. Duyn. Hunting for neuonal currents: absence of rapid MRI signal changes during visual evoked response. NeuroImage (in press).

Neuronal Dynamics…

J.P. Wikswo Jr et al. J Clin Neuronphy 8(2): 170-188, 1991

100 fT at on surface of skull

Magnetic field associated with single dendrite

Single dendrite having a diameter d, and length L behaves like a conductor with conductivity . Resistance is R=V/I, where R=4L/(d2 ). From Biot-Savart:

r2B= = 0

4Qr2

0

16d2 V

by substituting d = 4m, 0.25 -1 m-1, V = 10mV and r = 4cm

(typical measurement distance when using MEG)

the resulting value measured at the surface of a skull is:

B0.002 fT

J. Bodurka, P. A. Bandettini. Toward direct mapping of neuronal activity: MRI detection of ultra weak transient magnetic field changes. Magn. Reson. Med. 47: 1052-1058, (2002).

Because BMEG=100fT is measured by MEG on the scalp, at least 50,000 neurons (0.002

fT x 50,000 = 100 fT), must coherently act to generate such field. These bundles of neurons produce, within a typical voxel, 1 mm x 1 mm x 1 mm, a field of order:

MEGMEGMRI

MEGMEGMRI B

cm

cmB

r

rBB 1600

1.0

422

BMRI 0.2nT

Magnetic field associated with

bundle of dendrites

0.05 0.1 0.15 0.2 0.25 0.3

0.2

0.4

0.6

0.8

1

0.05 0.1 0.15 0.2 0.25 0.3

0.01

0.02

0.03

0.04

0.05 0.1 0.15 0.2 0.25 0.3

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Distance from source (cm)

∆B (nT)

∆ (Hz)

∆ (deg)

∆B = 100fT * (4cm/x)2

∆ = ∆B

∆= ∆ TE

Is 0.2 nT detectable?

wire

wire

X

Z

Current Phantom Experiment

calculated Bc || B0

B0

Measurement

Correlation image

Single shot GE EPI

70 A current

200

J. Bodurka, P. A. Bandettini. Toward direct mapping of neuronal activity: MRI detection of ultra weak transient magnetic field changes, Magn. Reson. Med. 47: 1052-1058, (2002).

Figure 1

B0<2nT

X=0

I=10 A, =40ms

FF

T((

t))

0.0 0.1 0.2 0.3 0.4 0.5

B0<0.2nT X=12.5 mm

Z

X

BR<0.2nT

BR<2nT

J. Bodurka, P. A. Bandettini. Toward direct mapping of neuronal activity: MRI detection of ultra weak transient magnetic field changes, Magn. Reson. Med. 47: 1052-1058, (2002).

0.002 0.004 0.006 0.008 0.01

0.5

1

1.5

2

Distance (m)

∆B (nT)

Main issues/obstacles:

•The effect is small

•Artifactual changes (respiration, cardiac) are order of mag larger

•The effect itself depends on geometry (phase/magnitude)

•The timing of the effect is variable

•BOLD still ubiquitous…

Human Respiration

00

400

200

BB0

TIME=110 s

=90o

BR10 nT

0.0 0.1 0.2 0.3 0.4 0.5

2x103

frequency ( Hz )

Slice selection Spatial encoding

Slice selection Spatial encoding

GE

SE

900

900 1800

One strategy for removing low frequency changes…

Experiment (human respiration)

GEspoiled

Spectral images

T

R =

1.0

sec

GE SE

GETE=55ms

SETE=55ms

M

20ms

S1: Optimal temporal position for 180 pulse; net phase shift induce by EMF>0

S1 S2

S1: For this 180 pulse temporal position net phase shift induce by EMF is close to zero

M. Singh, Sensitivity of MR phase shift to detect evoked neuromagnetic fields inside the head.

IEEE Transactions on Nuclear Science. 41: 349-351, (1994).

The use of spin-echo to “tune” to transients..

What should we be detecting?

Phase or Magnitude?

Phase vs. Magnitude

∆ phase

∆ magnitude, ?∆ phase

? ∆ magnitude, ? ∆ phase

In Vitro Studies

in vitro model

Organotypic (no blood supply or hemoglobin traces) sections of newborn-rat somato-sensory Cortex, or somato-sensory Cortex & Basal Ganglia

• Size: in-plane:~1-2mm2, thickness: 60-100m• Neuronal Population: 10,000-100,000

• Spontaneous synchronized activity < 2Hz• Epileptiform activity• Spontaneous beta freq. activity (20-30Hz)• Network Activity Range: ~ 0.5-15V

Cortex

100 m

Plenz, D. et al. Neurosci 70(4): 861-924, 1996

Basal Ganglia

Multi-Electrode Array (MEA)EEG Recording

1 kHz sampling rate, 20 minutes 8x8 electrode configuration

Multi-Electrode Array EEG recording

~100ms Transient

Event

in vitro MR protocol

Imaging (3T) NMR (7T)

• Spin-Echo EchoPlanar Imaging • free induction decay (FID) acquisition

SE EPI image

FID

• voxel size: ~3x3x3 mm

• Sampling Rate :1 Hz (TR: 1sec)

• TE: 60 ms

• Readout :44 ms

• slab size: ~2x10x1mm

• Sampling Rate :10 Hz (TR: 100ms)

• TE : 30 ms

• Readout : 41 ms

Six Experiments

two conditions per experiment

in vitro MR experiment design

Active : 10 min (600 images) neuronal activity present

Inactive : 10 min (600 images) neuronal activity terminated via

TTX administration

Imaging (3T) NMR (7T)

Six Experiments

two conditions per experiment

Active : ~17 min (10,000 images) neuronal activity present

Inactive : ~17 min (10,000 images) neuronal activity terminated via

TTX administration

Pre- and Post- MR scan electrical recordings

7 Tesla data

Power decrease between PRE & TTX EEG : ~ 81%

Decrease between PRE & TTX MR phase: ~ 70%

Decrease between PRE & TTX MR magnitude: ~ 8%

3 Tesla data

A: 0.15 Hz activity, on/off frequencyB: activityC: scanner noise (cooling-pump)

Active condition: black lineInactive condition: red line

A

B

C

1: culture

Hz

A

B

C

2: ACSF

Hz

1 2

ACSFCulture

0.15Hz map

FSE image

Strategies for Detection in Humans

•Time shifted sampling•Under sampling

Time shifted sampling

M. Singh, Sensitivity of MR phase shift to detect evoked neuromagnetic fields inside the head.IEEE Transactions on Nuclear Science. 41: 349-351, (1994).

J. Xiong, P. T. Fox, J.-H. Gao, Direct MRI Mapping of neuronal activity. Human Brain Mapping, 20: 41-49, (2003)

J. Xiong, P. T. Fox, J.-H. Gao, Direct MRI Mapping of neuronal activity. Human Brain Mapping, 20: 41-49, (2003)

R. Chu, J. A. de Zwart, P. van Gelderen, M. Fukunaga, P. Kellman, T. Hollroyd, J. Duyn. Hunting for neuonal currents: absence of rapid MRI signal changes during visual evoked response. NeuroImage, (in press).

Photodiode

MEG

8 Hz alternating checkerboard

Undersampling

TR

Alternating Checkerboard Frequency

Undersampling

Comparison of phase and magnitude of the MR signal in measuring neuronal activity [for Petes’ sake1,2]

James M. Kilner, Klaas E. Stephan, Oliver Josephs, Karl J. Friston

Wellcome Department of Imaging Neuroscience, 12 Queen Square, London

BOLD

Phase

Mag

0.5 Hz 0.5 Hz

Power spectraEyes closed Eyes open

Phase 0.12Hz

Closed Open

Magnitude 0.12 Hz

The principle and application of the Lorentz Effect Imaging

Song et al ISMRM 2000, p. 54

Other Methods??

iF

F

B

Lorentz Force

iF

F

B

Lorentz Force

-G G

Bo

1 k0.1 – 0.5 V

MRI Scanner Trigger Pulse

Experimental Setup: Phantom Study

Image Acquisition

Schematic illustration of the data acquisition.

encoding decoding

Rapid-flashing checkerboard

Optic Chiasm

EPI images without visual stimuli (A) and with visual stimuli (B).

rest active

A B

Preliminary Experiment on the Optical Nerve

•Need to rule out BOLD or other mechanisms

•Noise is larger than effect•MR sampling rate is slow•Neuronal activation timing is variable and unspecified•Models describing spatial distribution of ∆B across spatial scales are crude…could be off by up to an order of magnitude.

•We are understanding more about precise effects of stimuli.•“Transient-tuned” pulse sequences (spin-echo, multi-echo)•Sensitivity and/or resolution improvements•Lower field strengths? (effect not Bo dependent)•Simultaneous electrophysiology – animal models?•Synchronization improvements.•Again..models describing spatial distribution of ∆B across spatial scales are crude…could be off by up to an order of magnitude.

Caution…

Despair…

Hope…

EEG re-sampled using:TR: 1sec

TE: 60ms (spin-echo) Readout :44ms

Field calculations withvoxel volume: 27mm3

Bo : 3T

EEG re-sampled using:TR: 100msTE:30ms

Readout : 41ms(entire FID)

Field calculations with voxel volume: 20mm3

Bo : 7T

MR Simulations based on EEG

dB= Bneuronal(from EEG) dVdt; t:82ms, V=27mm3

ΔB

(n

Tes

la)

ΔB

(n

Tes

la)

dB= Bneuronal(from EEG) dVdt; t:71ms, V=20mm3

Assuming 100,000 neurons * 1.5nA per neuron

Simulated MR spectrumcalculated using sliding window FFT. Simulation based on previously shown electrical recording data

dB= µo It (neuronal from EEG) V / 2 π D dV dt ; t: (TE+Readout) ms, V=(voxel volume)mm3

Where:

V=sqrt(x2+y2+z2), where x,y,z are the voxel dimensionsD= the diameter of the culture (0.5 mm)µo= magnetic permeability of free spaceIt = the current produced from the neuronal network at a given point in time, t this is calculated from the EEG as follows: maximum current possible for the given culture (here 100,000 neurons * 1.5nA per neuron - reference: Nicholls JG, Martin AR, Wallace BG. From Neuron to Brain.

3rd ed. Sinauer, 1992.) multiplied by the following weighting function ( normalized EEG): resulting fields are consistent with those expected

(checked with Dietmar and literature – will give you ref. papers)