Y P O C Computational Modeling to T O Understand tDCS and tACS - - PowerPoint PPT Presentation

Computational Modeling to Understand tDCS and tACS

Flavio Frohlich

University of North Carolina - Chapel Hill

Department of Psychiatry Department of Cell Biology and Physiology Department of Biomedical Engineering Department of Neurology Neuroscience Center

www.facebook.com/FrohlichLabUNC

P L E A S E D O N O T C O P Y

UNC has a patent on (feedback) (t)ACS for modulation of cortical

• scillations as a therapeutic
• intervention. FF is lead inventor (US

PR App 61/899,954). UNC has determined the absence of a conflict

• f interest (COI) for the work

presented here and has determined a “COI with administrative considerations” for the clinical trials in the Frohlich Lab due to the use of hardware designed in the Frohlich Lab.

• We are working on our own

device hardware.

• I am writing a textbook

“Network Neuroscience” for Elsevier.

• I speak with people from

Neosync and Tal Medical and have received travel support.

• We use Neuroconn devices.
• My preferred brain stimulation

modality is espresso.

COI

P L E A S E D O N O T C O P Y

If your tDCS/tACS study only uses behavioral outcomes, either (1) you hit the jackpot and your original hypothesis got confirmed

• r (2) your results disagree with

your hypothesis, so you ???

Sellers et al. 2015

P L E A S E D O N O T C O P Y

VERTICAL INTEGRATION

Patients Model Systems

COMPLEXITY TRACTABILITY Clinical Trials Brain Stimulation, Human Neurophysiology In vivo (Animal) Electrophysiology In vitro (Animal) Electrophysiology Computer Simulations

P L E A S E D O N O T C O P Y

TRANSCRANIAL CURRENT STIMULATION STUDY DESIGN

Behavioral Target Network Target Target Engagement

P L E A S E D O N O T C O P Y

TARGET ENGAGEMENT

How do we best engage a network target? We need to understand what the effect of stimulation is on the brain in terms of neurophysiology.

P L E A S E D O N O T C O P Y

OUTLINE

• 1. Cellular Effects
• 2. Spatial Targeting
• 3. Targeting Network Dynamics

P L E A S E D O N O T C O P Y

ELECTRIC FIELDS

How do electric fields change electric signaling in neurons?

P L E A S E D O N O T C O P Y

“Anodal” Depolarized Soma Hyperpolarized Dendrite “Cathodal” Hyperpolarized Soma Depolarized Dendrite

P L E A S E D O N O T C O P Y

CABLE EQUATION

Frohlich and McCormick. 2010

P L E A S E D O N O T C O P Y

NEURONAL MORPHOLOGY AND STATE

Change in somatic membrane voltage:

• Increases with cable length.
• Decreases with membrane conductance.
• Increases with cable diameter.

A B vs.

Radmann et al. 2009

P L E A S E D O N O T C O P Y

Change in somatic membrane voltage can be modeled as a sub- threshold somatic current injection.

Frohlich and McCormick. 2010

P L E A S E D O N O T C O P Y

SUMMARY CELLULAR EFFECTS

Weak electric fields change the membrane voltage of neurons. The effect on an individual neuron depends on the field, the neuron, and the spatial relationship between the two. To study network-level effects, an adjusted somatic current injection can be used in computational models.

P L E A S E D O N O T C O P Y

SPATIAL TARGETING

• The electric field caused by current application

is a function of the electric conductivity of the tissue.

• Mathematically, the so-called Laplace equation

is numerically solved to determine the electric potential from the current application.

• The current application is modeled as a

boundary condition.

• The key parameter is the conductivity that

greatly differs between tissues.

• Current flow the strongest in skin and

cerebrospinal fluid (shunting).

Tissue Resistivity [Ohm cm] Copper 2e-6 CSF 64 Cortex 350 White Matter 650 Bone 8,000-16,000

P L E A S E D O N O T C O P Y

IMPLEMENTATION

• MR Scan
• Tissue segmentation
• Numerical solution based on dividing head in a

large number of small compartments (e.g. finite elements).

• 1. Develop you own code, typically using MR

analysis tools and a physics simulator.

• 2. Collaborate with groups that developed such a

tool.

• 3. Buy tool.

P L E A S E D O N O T C O P Y

P L E A S E D O N O T C O P Y

P L E A S E D O N O T C O P Y

Modeling performed by Angel Peterchev Sellers et al 2015

P L E A S E D O N O T C O P Y

P L E A S E D O N O T C O P Y

SUMMARY: SPATIAL TARGETING

• MR scan + Segmentation + EF

modeling = Spatial Targeting

• Conventional electrodes (scale: cm)

cause relatively broad electric field distributions.

• Electric fields are not only superficial.
• Smaller (and more) electrodes may

provide better spatial targeting.

P L E A S E D O N O T C O P Y

STRUCTURE DYNAMICS BEHAVIOR

P L E A S E D O N O T C O P Y

MODELING DYNAMICS

Frohlich 2014

P L E A S E D O N O T C O P Y

OSCILLATIONS

Caution: Most tACS literature refers to the peak-to-peak amplitude as amplitude.

P L E A S E D O N O T C O P Y

NETWORK DYNAMICS

• 1. Raw trace.
• 2. Spectrum: Power as a

function of frequency.

• 3. Spectrogram: Spectrum as

a function of time.

• 4. Coherence: Interaction

between two sites as a function of frequency.

Raw Trace Spectrum

P L E A S E D O N O T C O P Y

• 1. Raw trace.
• 2. Spectrum: Power as a function of frequency.
• 3. Spectrogram: Spectrum as a function of time.

Raw Trace Spectrogram

P L E A S E D O N O T C O P Y

• 1. Raw trace.
• 2. Spectrum: Power as a function of frequency.
• 3. Spectrogram: Spectrum as a function of time.
• 4. Coherence: Interaction between two sites as a function
• f frequency.

P L E A S E D O N O T C O P Y

TARGETING BRAIN NETWORK DYNAMICS

Write / Input tACS Transcranial Alternating Current Stimulation (tACS)

Neuroconn

Read / Output EEG

Berger 1929

P L E A S E D O N O T C O P Y

NATURALISTIC ELECTRIC FIELDS

Frohlich and McCormick. 2010

P L E A S E D O N O T C O P Y

PHASE SYNCHRONIZATION

Detuning: Difference between natural (endogenous) and stimulation (external) oscillation frequency.

P L E A S E D O N O T C O P Y

PHASE SYNCHRONIZATION

P L E A S E D O N O T C O P Y

ARNOLD TONGUE

Frohlich 2014

P L E A S E D O N O T C O P Y

SPIKING NEURAL MODEL (NETWORK)

Ali et al. 2013

P L E A S E D O N O T C O P Y

SPATIO-TEMPORAL DYNAMICS

Ali et al. 2013

P L E A S E D O N O T C O P Y

Ali et al. 2013

P L E A S E D O N O T C O P Y

STIMULATION PHASE

Ali et al. 2013

P L E A S E D O N O T C O P Y

HOTSPOTS

Ali et al. 2013

P L E A S E D O N O T C O P Y

NETWORK-LEVEL MECHANISM

Ali et al. 2013

P L E A S E D O N O T C O P Y

CELLULAR-LEVEL MECHANISM

Ali et al. 2013

P L E A S E D O N O T C O P Y

TARGETING A SUBPOPULATION

Ali et al. 2013

P L E A S E D O N O T C O P Y

NETWORK RESONANCE

Ali et al. 2013

P L E A S E D O N O T C O P Y

PHASE SLIPPING

Ali et al. 2013

P L E A S E D O N O T C O P Y

INTERACTING NETWORKS

Kutchko and Frohlich 2013

P L E A S E D O N O T C O P Y

MULTISTABILITY

“Rapid Fire” “Slow Propagating” “Spiral Waves”

Kutchko and Frohlich 2013

P L E A S E D O N O T C O P Y

STATE SWITCHING BY tACS

Kutchko and Frohlich 2013

P L E A S E D O N O T C O P Y

TARGET: ALPHA OSCILLATIONS

• Neurofeedback, rTMS (10 Hz), tACS
• f visual cortex to modulate

perception, Neosync, etc.)

• “Offline” state, long-range

functional connectivity, gating.

P L E A S E D O N O T C O P Y

THALAMUS: ALPHA, GAMMA, SPINDLES

Awake (“online”) Gamma Oscillations Awake (“offline”) Alpha Oscillations NON-REM sleep Spindles

P L E A S E D O N O T C O P Y

COGNITIVE ENHANCEMENT

High Creative Ideation Low Creative Ideation

“increased alpha power during creative ideation is among the most consistent findings in neuroscientific research on creativity” (Fink and Benedek, 2010)

Lustenberger et al. (2015)

P L E A S E D O N O T C O P Y

ENHANCING CREATIVITY

• Blinding was successful (p > 0.2).
• 10 Hz tACS significantly enhances creativity as measured by the Torrance

Test of Creative Thinking (7.45 % ± 3.11 % S.E.M.; F1,16 = 5.14, p = 0.036).

• No enhancement with 40Hz-tACS..

Lustenberger et al. (2015)

P L E A S E D O N O T C O P Y

STIMULATION ARTIFACT SUPPRESSION

Signal contaminated by stimulation artifacts Cleaned signal after artifact suppression

Artifact

Spectra showing peaks corresponding to artifacts Spectra of cleaned signal shows elimination of peaks

P L E A S E D O N O T C O P Y

OSCILLATION ENHANCEMENT

P L E A S E D O N O T C O P Y

STATE-DEPENDENT MODULATION

“Eyes Closed” “Eyes Open”

P L E A S E D O N O T C O P Y

STATE-DEPENDENT MODULATION

P L E A S E D O N O T C O P Y

P L E A S E D O N O T C O P Y

FEEDBACK tACS TO MODULATE SLEEP SPINDLES

P L E A S E D O N O T C O P Y

IMPROVING MEMORY CONSOLIDATION

P L E A S E D O N O T C O P Y

TARGET ENGAGEMENT

P L E A S E D O N O T C O P Y

SUMMARY: TARGETING NETWORK DYNAMICS

• Oscillations represent fundamental activity structure.
• tACS ideal to target cortical oscillations.
• Endogenous network dynamics represent oscillator to

be modulated by weak periodic perturbations.

• Arnold Tongue: Necessity of individualizing

stimulation frequency?

• Multistable dynamics: State-dependent stimulation

effects.

P L E A S E D O N O T C O P Y

Current Lab Members

Stephen Schmidt Charles Zhou Kristin Sellers Michael Boyle Caroline Lustenberger Chunxiu Yu Sankar Alagapan Apoorva Iyengar Yuhui Li Guoshi Li Juliann Mellin Courtney Lugo Philipp Lustenberger

Alumni Lab Members

Mohsin Ali Katrina Kutchko

Collaborators

ECOG: Dr. Haewon Shin Sleep Spindles: Dr. Bradley Vaughn Modeling ECOG: Dr. Jeremy Lefebvre Electric Field Spatial Targeting: Dr. Angel Peterchev SCZ Clinical Trial: Dr. Fred Jarskog, Dr. John Gilmore Mood Disorders Clinical Trials: Dr. David Rubinow

Funding

NIMH BRAINS R01 MH101547, NIMH R21MH105557, NIMH R21MH105574, Human Frontier Science Program, UNC School of Medicine, Department of Psychiatry, NCTraCS (CTSA #1UL1TR001111), Foundation of Hope, UNC SOM TTSA, NARSAD, Patient Donations.

P L E A S E D O N O T C O P Y

Thank you for your attention. Feedback and/or questions: flavio_frohlich@med.unc.edu

P L E A S E D O N O T C O P Y