SDSU Foundations of Neuroimaging (PSY0569, PSY0769)

FALL 2020

Course Flyer:

Marty Sereno — email: msereno - AT - sdsu
lecture recording times: Mon/Wed/Fri 9:00-9:50 AM (grad session: Fri 8:00-8:50 AM)
lecture recording location: SSW 2667 (Learning Glass Studio, Student Services West)
All lectures live-streamed on Zoom

Learning Glass lecture recordings
Sereno Lecture Notes (123-page PDF [19MB] — single-page links below, last update: 11 Dec 2020)
Huettel, S., A.W. Song, G. McCarthy (2014) Functional Magnetic Resonance Imaging, 3rd ed.
Additional background reading

Homework #0 (warm-up, no need to turn in)
Homework #1 (due 10/12/2020, paper printout, incl. code and graphs)
Homework #2 (due 11/23/2020, paper printout, incl. code and graphs, brain image is here)
Final Paper: (ugrad=5pg, grad=10pg) literature review on narrow methodological topic (start search in
    Magnetic Resonance in Medicine, Neuroimage, Human Brain Mapping)

Learning Objectives:
Students will be able to do the following:
   (1) explain precession/excitation/recording/contrast of magnetic resonance signals and echoes using the Bloch equation
   (2) compute Fourier transform, use to explain RF stim, gradients, signals generate k-space data, how recon. works
   (3) diagram main classes anatomical/functional pulse sequences
   (4) describe diffusion, perfusion, and spectroscopic imaging
   (5) describe origin/localization of EEG/MEG signals, cortical surface-based methods, and how to combine w/fMRI
   N.B.: consult with me if a disability hinders your performance so we can use University resources to maximize learning

Description and Prerequisites:
This course covers the physical and mathematical foundations of structural, functional, diffusion, and perfusion MRI, fMRI time series analysis, cortical-surface based reconstruction and data analysis, and the neural basis, recording, and localization of EEG and MEG signals. Although this course does not assume a background in linear algebra, vector calculus, differential equations, electromagnetism, the Fourier transform, and convolution, students will be expected to develop a solid grasp of a number of key equations underlying the different neuroimaging methods and solve simple Matlab problems. We will go slower than a typical engineering class and there will be plenty of time for questions. Here are two recently taught neuroimaging courses at UCSD for reference. Tom Liu's Bioengineering 280A — Principles of Biomedical Imaging (Fall 2015), provides a stronger mathematical foundation in the fundamentals of the Fourier transform and linear systems theory and covers ultrasound and CT in addition to MRI. David Dubowitz and Rick Buxton teach a two-quarter SOMI 276 — School of Medicine fMRI Course (2016/2017) in the Radiology Department.

Lecture Topics — Fall 2020 (pdf)

Week of Aug 24 (Mon/Wed/Fri)Introduction

  • Introduction to Neuroimaging — MRI, fMRI, EEG, MEG
  • MRI hardware
  • Spin and Precession

    Week of Aug 31 (Mon/Wed/Fri)Bloch Equation

  • Dot/Cross/Complex Products
  • Bloch Equation
  • Precession Solution
  • Effects of Change (M, B, Angle) On Precession Freq
  • Initial-Value Solution to T2 Differential Equation
  • Verify T1 Re-Growth Solution
  • Simple Matrix Operations
  • Bloch Equation, Solution — Matrix Version
  • Bloch Equation: Excitation In Rotating Frame
  • Bloch Equation: Summary

    Week of Sep 07 (Wed/Fri)Signal Equation

  • [no class Mon, Sep 07]
  • RF Excitation
  • Signal Equation
  • Phase-Sensitive Detection

    Week of Sep 14 (Mon/Wed/Fri)Echoes

  • Free Induction Decay
  • Spin Echo
  • Spin Echo Equations
  • Stimulated Echo, Spin Echo Trains
  • Extended Phase Graphs
  • 3-Pulse Echo Amplitudes
  • HyperEcho Train
  • Gradient Echo, Gradient Echo Trains

    Week of Sep 21 (Mon/Wed/Fri)Using the Bloch Equation

  • Saturation-Recovery Signal
  • Imperfect 90 deg Approach to State State
  • Inversion Recovery Signal
  • Spin Echo Signal
  • Gradient Echo Signal
  • Generalized MDEFT Signal
  • Magnetization Transfer Contrast
  • SNR/tSNR/CNR (e.g., Gray/White)
  • Signal-to-Noise

    Week of Sep 28 (Mon/Wed/Fri)Fourier transform

  • Complex Algebra
  • Fourier Transform
  • Negative Exponents, Orthogonality
  • Inverse Fourier as Correlation w/Cos,Sin
  • Spatial Frequency Space (k-Space) — 3 Equivalent Representations
  • One k-Space Point
  • Simple Image (1 Freq), Shifted Simple Image
  • Center of k-Space, Complex Image
  • Kx plus Ky Rotates, Special Case Sum of Reals

    Week of Oct 05 (Mon/Wed/Fri)Gradients, Slice Selection

  • Gradient Fields
  • Gradient Combination
  • Slice Selection
  • RF Pulse Details
  • Slice Select RF Pulses
  • Introduction to Frequency-Encoding —It's A Misnomer
  • Frequency-Encoding—Avoid This Intuition
  • Frequency-Encoding—Correct Intuition (=Phase-Encoding)

    Week of Oct 12 (Mon/Wed/Fri)MRI Image Formation

  • 1st Take-Home Due
  • Imaging Equation (1D)
  • Phase Encoding
  • 3D Imaging (2nd Phase-Encode)
  • Spin Phase in Image Space
  • Gradients Move Signal in k-Space

    Week of Oct 19 (Mon/Wed/Fri)Image Reconstruction

  • Image Space and k-space (311K MPEG, R.-S. Huang)
  • Image Reconstruction
  • Aliasing and FOV
  • Under/Over Sample
  • Replicas, FTs
  • Point-Spread Function
  • General Linear Inverse for MRI Reconstruction

    Week of Oct 26 (Mon/Wed/Fri)Practical Pulse Sequences

  • Fast Spin Echo
  • 3D Multi-Slab Fast Spin Echo
  • 3D Single-Slab Fast Spin Echo (SPACE)
  • Fast Gradient Echo
  • Quantitative T1 (Intro, Other Methods)
  • Quantitative T1 (2-Flip-Angle Method)
  • Gradient Echo EPI
  • Spin Echo Size Selectivity
  • Spin Echo EPI
  • Coil Fall-Off, Undersampling, GRAPPA, SENSE
  • Simultaneous Multi-Slice (blipped CAIPI)
  • SMS cont.: slice-GRAPPA, split-slice-GRAPPA
  • 3D Echo Volume Imaging
  • Single-Shot Spiral
  • 3D Spiral Inversion-recovery FSE

    Week of Nov 02 (Mon/Wed/Fri)Image Artifacts

  • Fourier Shift Artifacts
  • EPI vs. Spiral Artifacts
  • Image-space View Localized B0 Defect
  • Effect Local B0 Defect on Reconstruction
  • Shimming and B0-Mapping
  • Gradient Non-linearities
  • Motion-detection with Navigator Echoes
  • RF Field Inhomogeneities

    Week of Nov 09 (Mon/Fri)Diffusion, Perfusion, and Spectroscopy

  • Diffusion-Weighted Imaging and Tract Tracing
  • Practical Diffusion-Weighted Sequences
  • Perfusion Imaging (Arterial Spin Labeling)
  • pCASL
  • Off-Resonance RF Excitation
  • [no class Wed, Nov 11]
  • Combining Spectroscopy with Imaging

    Week of Nov 16 (Mon/Wed/Fri)Block Design, Phase-Encoded Design

  • Phase-Encoded Stimulus and Mapping
  • Convolution
  • General Linear Model and Solution
  • General Linear Model — Geometric Picture
  • Cluster Correction — 3D and Surface-Based
  • Why Use Surfaces?
  • Normalize, Strip Skull
  • Spring Force in Detail
  • Non-isotropic Filter, Region-Growing
  • Tessellation: 3D -> 2D
  • Energy Functional for Smooth/Inflate/Final
  • Cortical Unfolding and Flattening
  • Automatic Topology Repair
  • Sulcus-Based Alignment

    Week of Nov 23 (Mon-only)Cortical Surface Based methods

  • 2nd Take-Home Due
  • Cortical Thickness Measurement
  • Mapping Cortical Visual Areas
  • [no class Wed/Fri, Nov 25/27]

    Week of Nov 30 (Mon/Wed/Fri)Source of EEG/MEG

  • Neural Source of EEG/MEG
  • Intracortical Circuits and EEG/MEG
  • Grad, Div, Curl
  • 1D/2D/3D Current Source Density
  • 1D/2D Current Source Density Experiments
  • Intracortical Basis of EEG
  • Maxwell Equations — Low Frequency Limit
  • Why We Can Ignore Magnetic Induction
  • Monopole, Dipole Current Sources

    Week of Dec 07 (Mon/Wed)Neuroimaging EEG/MEG

  • Forward Solution — Analytic, Boundary Element (incomplete)
  • Matrix Formulation of the Linear Forward Solution
  • Why Localize?
  • Derivation of Ill-Posed Minimum Norm Linear Inverse
  • Minimum Norm: Why It Appropriately 'Devalues' Deeper Sources
  • Two Equivalent Formulations — One Is Easier to Compute
  • Surface-Normal Constraint and its Problems
  • fMRI-weighted Inverse Solutions
  • Noise-Sensitivity Normalization of Mininum Norm
  • Noise-Sensitivity Normalization — Intuition
  • Point-Spread/Crosstalk, Conclusions
  • Using Spatiotemporal Covariance of Sensors (MUSIC)
  • Sensor Covariance: Calculate Weights
  • Sensor Covariance: Insert Weights into Inverse
  • Sensor Covariance: How it Helps
  • Non-linear Fitting of Moveable Dipoles
  • [no class Fri, Dec 11]

    Week of Dec 16 Final Paper/Exam

    Final paper/exam due Dec 17

    website last modified: 03 Aug 2020
    Scanned/video'd class notes (pdf, links above) © 2020 Martin I. Sereno
    Supported by NSF 0224321, NIH MH081990, Royal Society Wolfson