Functional Overview & Biophysical Principles
MicBrain provides an interactive, three-dimensional anatomical platform demonstrating the mapping of frequency-dependent acoustic dynamics onto corresponding functional regions of the human cerebral cortex. Built upon spatial meshes from the Destrieux anatomical cortical atlas, the visualizer decomposes incoming real-time audio data into fifteen discrete functional signal domains. These domains reflect local cortical field potential dynamics, offering a physiological bridge between acoustic signals and functional brain activity.
In biological neuroanatomy, acoustic stimuli captured by the cochlea undergo neural transduction along the auditory pathway. Signals pass through the spiral ganglion cells of the auditory nerve, traverse the cochlear nuclei and superior olivary complex within the brainstem, project through the inferior colliculus, and synapse at the medial geniculate body (MGB) of the thalamus. Thalamocortical projections subsequently route acoustic representations to Heschl’s gyrus (transverse temporal gyrus, primary auditory cortex $A1$).
MicBrain replicates this functional partitioning by mathematically mapping specific acoustic frequency sub-bands $B_k \in [0, 14]$ into primary, secondary, and higher-order cortical regions. Lower frequencies correlate with deep midline limbic and frontal structures, mid-range vocal frequencies target primary auditory and temporal association sulci, and high-frequency components map to dorsal parietal integration nodes and occipital visual regions.
Instrument Calibration & Workflow Instructions
Operate the laboratory visualizer through either live environmental microphone streams or automated synthetic physiological signal sweeps:
1. Signal Input Selection: Click Start Microphone to request local media permissions and capture physical ambient noise, or click Start Demo at the top of the control sidebar to execute multi-frequency synthetic formants and rhythmic neural sweeps.
2. 3D Camera Navigation: Rotate the anatomical model using a primary click-and-drag gesture. Pan the focal point via secondary click-and-drag, and adjust vertical and radial zoom levels with mouse wheel scrolls or touch pinches.
3. Active Regions Selection: Adjust the Active Regions slider to dynamically limit the display from a sparse subset of highly excited anatomical parcellations up to all 150 Destrieux cortical structures.
4. Input Gain & Noise Gate Calibration: Increase Input Gain to boost weak environmental audio signals. Raise the Noise Gate Threshold to filter out low-level ambient noise, ensuring baseline signals do not trigger false cortical activations.
5. Adaptive AGC & DMN Dynamics: Toggle Spectral Whitening (AGC) to dynamically equalize input frequency bands, preventing low-frequency dominance. Adjust Resting-State Waves (DMN) to inject low-frequency sinusoidal oscillations simulating metabolic fluctuations in the Default Mode Network.
Architectural Engineering & Signal Equations
Auditory signals are captured via the browser Web Audio API using an AnalyserNode configured with a Fast Fourier Transform ($FFT$) window size $N = 2048$. The raw frequency spectrum is discretized across normalized Fourier bin indices $k \in [0, N/2 - 1]$. The energy within a target frequency slice $I_b = [k_{\min}, k_{\max}]$ is calculated using discrete bin aggregation:
$$P_b = \frac{1}{|I_b|} \sum_{k \in I_b} |X[k]|^2$$
To enforce dynamic spectral balance across non-linear human auditory bandwidths, Adaptive Spectral Whitening normalizes energy levels using dynamic rolling peak $P_{\text{peak},b}(t)$ and noise floor $P_{\text{floor},b}(t)$ tracking equations:
$$Y_b(t) = \frac{P_b(t) \cdot G - P_{\text{floor},b}(t)}{P_{\text{peak},b}(t) - P_{\text{floor},b}(t) + \epsilon}$$
Where $G$ represents user-defined Input Gain, and $\epsilon = 10^{-6}$ prevents division by zero. The resulting response is smoothed across time frames using an exponential moving average filter with smoothing factor $\alpha = 0.08$:
$$E_b(t) = (1 - \alpha) \cdot E_b(t - \Delta t) + \alpha \cdot \max\left(0, \frac{Y_b(t) - \theta_{\text{gate}}}{1 - \theta_{\text{gate}}}\right)$$
To simulate intrinsic physiological metabolic fluctuations, each Destrieux mesh region $R_j$ receives a modulated visual excitation score $S_j(t)$ combining spectral energy with slow-wave resting state oscillations:
$$S_j(t) = E_{B(R_j)}(t) \cdot \left[1.0 + M_{\text{DMN}} \cdot \sin(\omega_j \cdot t + \phi_j)\right]$$
Where $M_{\text{DMN}}$ is the resting-state modulation depth, while $\omega_j$ and $\phi_j$ represent unique low-frequency spatial phase parameters derived deterministically from the region name hash.
Future Directions & Clinical Diagnostic Roadmaps
The prospective technical roadmap for MicBrain focuses on integrating high-density Electroencephalography ($EEG$) and Magnetoencephalography ($MEG$) source-reconstruction models directly onto the 3D Destrieux anatomical meshes. Incorporating standardized Boundary Element Models ($BEM$) and finite element head models ($FEM$) will transform this visualizer from a conceptual neuro-acoustic tool into a real-time clinical diagnostic workspace.
Subsequent milestones will introduce automated anatomical label overlays detailing Brodmann areas ($BA41/42$, $BA4$, $BA17$), alongside dynamic functional network coloring for resting-state subnetworks (Salience, Central Executive, and Default Mode Networks).
Explore Related Laboratory Environments