← Back to Course Home

Interactive Neuroimaging Demos

A suite of interactive web-based simulations designed for NBL 425/625 to explore the core physical principles, spatial encoding schemes, contrast mechanisms, statistical modeling, and artifact origins in human neuroimaging.

Course Demos

• MRI Simulator: Scanner, Sequences, k-space & Spectroscopy

  Integrated 3 T Simulator · Pulse-Sequence Language, Phantoms, Artifacts, Reconstruction & MRS

  The course's integrated scanner. Write pulse sequences in a small language, image a brain or a test phantom on a schematic 3 T scanner, and watch k-space fill line by line as the hardware plays the sequence. Reconstruct the data several ways (filters, zero-filling, partial Fourier, SENSE, gridding), switch on artifacts that arise from their physical cause, and acquire a non-localised proton FID with water suppression and hands-on shimming. Thirteen guided lessons.

• EPI Geometric Distortion & Susceptibility Dropout

  EPI Artifacts · Field Offsets ΔB0, Bandwidth & TOPUP Unwarping

  Visualizes how air-tissue interfaces (paranasal sinuses) interact with the low phase-encoding bandwidth of Echo Planar Imaging to cause geometric stretching (P→A) vs. compression (A→P) and intra-voxel signal blackout, and how reversed phase-encode calibration (TOPUP) restores geometry.

• fMRI GLM & Convolution Studio

  fMRI Analysis · HRF Convolution, Design Matrices, Prewhitening & OLS Regression

  Design cognitive timelines (block vs. event-related), convolve impulses with the canonical double-gamma Hemodynamic Response Function, inspect collinearity and variance inflation (VIF), and solve the General Linear Model ($Y = X\beta + \epsilon$) in real time with statistical contrasts ($t$-tests). Colour the noise with physiological oscillations and AR(1) autocorrelation, switch on prewhitening, and audit the null false-positive rate to see how far ordinary least squares over-states its own significance when the residuals are not independent.

• fMRI Preprocessing Pipeline Workbench

  fMRI Analysis · Realignment, Slice Timing, Smoothing & High-Pass Filtering

  Acquire a synthetic run corrupted by head motion, interleaved slice timing, drift and spatially correlated physiological noise, then switch each preprocessing stage on and off and watch the statistics respond. Realignment parameters are genuinely estimated by image registration rather than read back from the simulation, so the residual error is real. Two activation foci of different size expose the matched-filter trade in spatial smoothing, including the point where a larger kernel fuses them into one.

• Whole-Brain Thresholding & Multiple Comparisons

  Statistical Inference · Uncorrected, Bonferroni, FDR & Cluster Extent

  Run a real voxelwise GLM across a whole slice and threshold the resulting $t$-map four ways side by side. The cluster-extent threshold is calibrated by Monte Carlo against null data rather than assumed, and a built-in audit measures each method's family-wise error rate by analysing 200 datasets containing nothing at all. Includes the Benjamini–Hochberg ladder plot and the dead-salmon demonstration.

• Group-Level Inference: Fixed vs. Random Effects

  Second-Level Analysis · Variance Components, Power & Generalisation

  Simulate a full group study, run both second-level analyses on identical data, and see why fixed effects cannot support a claim about people. A forest plot contrasts the two confidence intervals, Monte Carlo power curves show when scanning longer stops helping and recruiting is the only lever left, and a validity audit measures how often each analysis invents a population effect that is not there.

• Free Induction Decay (FID) & Spin Relaxation

  Induction Physics · Faraday's Law, Mxy Evolution, T1 Recovery & T2* Dephasing

  Directly connects the 3D time evolution of the net magnetic moment to the voltage signal induced in an RF receiver coil. Remembers longitudinal $M_z$ memory across pulses to demonstrate $T_1$ saturation, inversion recovery, and the null point.

• Gradient Encoding Visualizer

  Spatial Encoding · Frequency, Phase & Slice Selection

  Visualizes how linear magnetic field gradients make the Larmor precession frequency position-dependent, demonstrating slice selection (G_z), phase encoding blips (G_y), and frequency encoding readout (G_x).

• Pulse Sequence Timing Diagram

  Sequence Engineering · Spin Echo, TR & TE

  An interactive spin-echo timing diagram. Adjust TE to see the 180° refocusing pulse and echo move synchronously; adjust TR to explore repetition timing, dead time, and phase-encode step evolution.

• TR/TE Contrast Explorer

  Contrast Weighting · T1, T2 & Proton Density

  Manipulate TR and TE sliders to see real-time signal calculations across four tissue types (White Matter, Gray Matter, CSF, and Fat), dynamically identifying T1-weighted, T2-weighted, and PD-weighted regimes.

• k-space Explorer

  Fourier Reconstruction · Central vs. Peripheral Spatial Frequencies

  Filter k-space with a variable centered window to observe how the center carries coarse image contrast while the periphery contributes edge sharpness. Illustrates Gibbs truncation ringing artifacts.

• MRI Artifact Simulator

  Artifact Physics · Chemical Shift, Aliasing & Motion

  Introduce deliberate k-space and image distortions to inspect Chemical Shift (fat-water frequency displacement), Aliasing / Wrap-around (reduced FOV), and Motion Ghosting (phase-encode modulation) with variable severity.

• DTI Tensor & Tractography Explorer

  Diffusion · Eigenvalues, FA, Streamline Integration & Crossing Fibers

  Build a diffusion tensor from its eigenvalues and orientation, read off FA, MD and the Westin linearity / planarity / sphericity measures, and see the resulting signal attenuation across a 6-, 15- or 30-direction gradient scheme. A synthetic crossing then shows why deterministic tractography stops at fiber intersections, and why loosening the turning-angle limit to get through produces anatomically false tracts.

• fNIRS Modified Beer–Lambert Unmixer

  Optical Imaging · Extinction Spectra, Isosbestic Point & Matrix Conditioning

  Choose two laser wavelengths inside the near-infrared window and see how that single decision determines whether oxy- and deoxy-haemoglobin can be separated at all. Includes the measured extinction spectra, the Modified Beer-Lambert forward model, and a simulated block-design recording solved back to concentrations, with the condition number of the 2x2 system reported live.

• fNIRS Simulator: From Photons to Haemoglobin

  Optical Imaging · Recording Language, Banana Sensitivity, Beer-Lambert, Confounds, Short-Channel Regression, GLM & HD-DOT

  Describe a recording in a small language (sources, detectors and short-separation channels over a 10-20 position, wavelengths, sampling rate and a block or event-related timeline with conditions) and record it from a five-layer head with an adjustable cortical activation, a connected pair, an infant, a thick skull or dark hair. Watch photons form the banana and trade depth against light as the separation grows, pick wavelengths against the HbO, HbR and water spectra, and follow intensity to optical density to concentration through the modified Beer-Lambert law. Switch the heartbeat, breathing, Mayer waves, a task-evoked scalp response, motion, a loose optode and room light on and off, then check quality, correct motion, filter, regress out the short channels, and watch the response emerge in the block average and the GLM while a false scalp activation disappears. Also compares fNIRS with BOLD fMRI and shows how a high-density array becomes diffuse optical tomography.

• PET Simulator

  Molecular Imaging · Protocol Language, Coincidences, Sinograms, FBP/OSEM/TOF, Corrections & Kinetic Modelling

  A teaching PET scanner in the browser. Write a scan protocol (tracer and isotope, dose, uptake, static or dynamic framing, energy and coincidence windows, time of flight), image a point source, cylinder, hot and cold spheres or a brain with FDG, amyloid, dopamine-transporter and water tracers in normal, Alzheimer's and Parkinson's conditions. Watch annihilations become true, scattered and random coincidences, inspect the sinogram, reconstruct with FBP or OSEM, switch attenuation, scatter, randoms, noise, positron range, motion, partial volume and a misaligned mu map on and off along with their corrections, and fit compartment models, SUV and SUVR against the true binding.

• PET Pharmacokinetics & 2-Tissue Compartment Simulator

  Molecular Imaging · Rate Constants, TACs & Patlak/Logan Plots

  Model intravenous radiotracer delivery across the blood-brain barrier into free and metabolically trapped or receptor-bound compartments. Evaluate real-time Time-Activity Curves and perform Patlak (irreversible FDG glucose) and Logan (reversible ligand) graphical linearizations.

• EEG Simulator: From Dipoles to ERPs

  Electrophysiology · Recording Language, Volume Conduction, Artifacts, ICA, ERPs & Dipole Fitting

  Describe a recording in a small language (cap, reference, sampling rate, filters and a stimulus timeline with trials and conditions), record it from dipole sources in a four-shell head, and switch blinks, eye movements, muscle, cardiac, line noise, drift, pops and a bad channel on and off. Then re-reference, filter with the distortions visible, remove ICA components checked against the ground truth, watch an ERP emerge as SNR grows with $\sqrt{N}$, and fit a dipole to see how noise, electrode count and depth make the inverse ambiguous.

• MEG Simulator: From Dipoles to Source Estimates

  Electrophysiology · Recording Language, Sensor Types, Shielding, SSP & SSS, ICA, Averaging & Inverse Methods

  Describe a recording in a small language (sensor array, sampling rate, filters, noise suppression and a stimulus timeline with trials and conditions) and record it from dipole sources in a spherical head with magnetometers, axial or planar gradiometers, or on-scalp OPMs. Compare the MEG field map with the EEG potential map of the same source to see the 90° rotation, the silence of radial and deep sources and the smearing of EEG by the skull. Switch environmental noise, line noise, the heart, blinks, muscle, dental work, SQUID jumps, a bad channel and head movement on and off, then the shielded room, gradiometers and reference sensors, SSP and SSS, filters, ICA and movement compensation. Watch an evoked field emerge as SNR grows with $\sqrt{N}$, coregister the head with an MRI, and compare dipole fitting, minimum norm and a beamformer on the same data.

• EEG Forward Problem & Scalp Potential Mapper

  Electrophysiology · 3-Shell Head Model, Volume Conduction & 10-20 Topomaps

  Position and orient equivalent current dipoles in a 3-shell concentric head model (Brain, high-resistivity Skull, Scalp). Observe how volume conduction through the resistive skull blurs focal cortical activity, and inspect the resulting 10-20 scalp electrode voltage distributions.

• EEG vs. MEG: Radial vs. Tangential Dipole Sensitivity

  Electromagnetism · Biot-Savart Law, Gyral vs. Sulcal Sources & Sensor Physics

  Explore why MEG is completely blind to radial dipoles on gyral crowns and selectively records tangential dipoles in sulcal banks, whereas EEG detects both. Includes dual side-by-side electric potential and magnetic flux density topomaps.

• Simultaneous EEG-fMRI Artifact Decontamination Workbench

  Multi-Modal Processing · Gradient Spikes, BCG Removal & ERP Recovery

  Step through an interactive multi-stage decontamination pipeline to strip away 20 millivolt imaging gradient switching bursts via Average Artifact Subtraction (AAS) and cardiac ballistocardiogram (BCG) artifacts via Optimal Basis Sets (OBS), unmasking the clean microvolt visual P100 evoked response.

• Structural Connectome Matrix & Graph Theory Studio

  Connectomics · Adjacency Matrix, Small-Worldness, Hubs & Targeted Attack Resilience

  Construct whole-brain structural connectivity matrices from streamline tractography. Calculate characteristic path length, clustering coefficient, small-worldness (σ), and test network attack vulnerability by comparing random axonal damage against targeted rich-club hub lesions.

• PET Sinogram & Tomographic Reconstruction Workbench

  Tomography · Coincidence LORs, Radon Sinograms, FBP & Iterative MLEM

  Map positron annihilation coincidence lines of response into Radon sinograms. Invert the intrinsic 1/r backprojection blurring with Ram-Lak ramp and Hann filters, and run iterative Poisson MLEM with 511 keV tissue attenuation correction.

• EEG Inverse Problem & Source Localization Studio

  Source Localization · Underdetermined Inverses, MNE, Depth Weighting & LCMV Beamformers

  Reconstruct cortical current dipoles from 16-channel scalp electric potentials. Compare unregularized pseudoinverse noise explosion against Tikhonov Minimum Norm Estimation (MNE), depth-weighted wMNE, and spatial LCMV beamforming.

• fNIRS Photon Transport & Short-Channel Scalp Regression

  Diffuse Optics · Monte Carlo Transport, Banana Paths & Scalp Mayer Wave Decontamination

  Simulate photon packet scattering random walks through layered scalp, skull, and brain cortex. Test dual-distance optodes to see how short-channel (8 mm) regression strips away confounding systemic blood pressure Mayer waves (0.1 Hz) to isolate true cortical activation.

• MRS Metabolite Basis Set & LCModel Spectral Fitter

  Spectroscopy · Chemical Shift, Metabolite Basis Vectors, Splines & CRLB Quantitation

  Deconstruct in-vivo 1H MR spectra into basis spectra for NAA, Creatine, Choline, Myo-inositol, Glutamate, and Lactate. Explore B0 field dispersion (1.5T vs 3T vs 7T), shim linewidths, macromolecular baselines, and Cramer-Rao Lower Bounds (CRLB %SD) across clinical tumor and stroke presets.

• Multi-Modal Fusion & Representational Similarity Analysis (RSA)

  Data Fusion · Second-Order Isomorphism, Millisecond Dynamics & Spatial Categorization

  Bridge millisecond EEG dynamics and millimeter fMRI spatial activations without forced source constraints. Track condition-by-condition Representational Dissimilarity Matrices (RDMs) across post-stimulus time to map the spatiotemporal hierarchy from V1 to LOC and FFA.

• Optically Pumped Magnetometers (OPM) & Wearable MEG Studio

  Emerging Quantum Modalities · SERF Physics, Standoff Distance, Pediatric Imaging & Motion Robustness

  Compare room-temperature wearable OPM arrays against rigid cryogenic SQUID helmets. Test how eliminating the 25-35 mm vacuum dewar gap yields a 4-5x boost in magnetic flux (B ∝ 1/r^2), enables pediatric imaging, and allows natural head motion with active background nulling coils.

• Brain-Wide Association Studies (BWAS) Power & Replication Simulator

  Big Data · Statistical Power, Winner's Curse Effect Inflation & Out-of-Sample Replication

  Simulate brain-behavior association studies across sample sizes from N = 25 to N = 4,000. Discover why modest true effects (|r| ≈ 0.10) produce severe effect size inflation (r_obs ≈ 0.50) in underpowered cohorts, and run 1,000-study Monte Carlo audits to see why thousands of scans are required for honest replication.

← Back to Course Home  ·  Full interactive index →  ·  View Future Demos Roadmap →