About this site

Electrical engineering and neuroscience have been trading ideas for a century, usually without much ceremony. Nearly everything we know about the nervous system arrived through an amplifier, an electrode, or a converter. And nervous systems, in return, keep turning out to be solutions to engineering problems that circuit designers are still working on.

This is a collection of those ideas: fundamental technologies, innovations, and concepts, mostly other people's, some of them mine. It currently holds 19 entries across 9 concepts.

How it's organized

Every entry declares which way the idea travelled:

• EE → Neuro — the instruments that made discoveries possible.
• Neuro → EE — biology read as a design manual.
• A few travelled both ways, and appear on both sides.

Cutting across that, concepts are where the two directions land on the same idea from opposite ends. Those pages are the most interesting ones here. There is also a timeline if you would rather read in the order things happened.

On maturity

Entries are marked seed, growing, or evergreen. A seed is a stub I wanted recorded before I forgot it — honest about being thin, rather than sitting unpublished for a year. Where an entry rests on someone's specific work, that work is cited at the bottom.

Where I come in: TMS

The corner of this border I actually work on is transcranial magnetic stimulation. A coil discharged over the scalp induces an electric field inside the skull without touching the brain — no electrodes, no surgery, and no current forced through the scalp and everything beneath it on the way to the target. That makes TMS unusual among neurostimulation methods in being both well established, with clinical use in treatment-resistant depression going back well over a decade, and — up to a point — interpretable. The field it induces is a physical quantity you can compute from Maxwell's equations and the geometry of a head, rather than a dose whose mechanism has to be inferred statistically.

The interpretability stops at the neuron. We know repetitive TMS treats depression largely because we observe that it does, and there is still no satisfying account of why one pulse shape or timing pattern should work better than another. That gap is the interesting part, because unlike a drug, a magnetic field has parameters an engineer can actually turn: amplitude, shape, direction, timing, repetition rate.

Historically you could not turn most of them. A conventional stimulator discharges a capacitor through the coil and produces whatever waveform the resulting LC ringdown happens to give, at whatever rate the coil's thermal budget allows — the shape was never a design choice, just a consequence of the circuit. My work is on removing that constraint with modular power electronics: a 2.8 kV, 6 kA modular stimulator that synthesises the pulse instead of inheriting it, so shape, timing, and repetition rate all become things you set. It is closer in spirit to a digital-to-analog converter than to a capacitor discharge, operating at a few megawatts.

Two pieces of that, in case either is useful to you:

• Energy-optimized monophasic pulses. Monophasic pulses are the ones neuroscientists tend to want and also the ones that heat the coil fastest, which is why they have been hard to use at repetitive rates. Giving a sinusoidal current a long, low pre-phase opposite to its short main phase produces an equivalent electric field in the brain while cutting the RMS current — and so the coil loss — by roughly three quarters.
• Modulation-enhanced nearest-level quantization. A modular converter has to keep choosing which discrete output level to switch to. Nearest-level modulation switches rarely but quantizes coarsely; carrier-based PWM does the opposite. Low-pass filtering the nearest-level residue before it reaches a one-bit PWM stage recovers most of both.

I do this in the Brain Stimulation Engineering Lab at Duke, with Stefan M. Goetz and Angel V. Peterchev. The machines themselves, with numbers, are under build.

My own entries here

Most of this site collects other people's work. These are mine, kept separate so the distinction is never in doubt.

Corrections

Genuinely welcome, especially on attribution — if I have credited the wrong people for something, I would rather know. My CV, publications, projects, and personal writing live at jinshui.me; this site stays on topic. The source is on GitHub, and my email is on jinshui.me.