🧠 neuroscience

Exploring the sensitivity limits of neuronal current imaging with MRI and MEG in the human brain

This study demonstrates that while MEG and BOLD-fMRI confirmed robust neuronal and hemodynamic activation during visual stimulation, spin-lock fMRI failed to reliably detect direct neuronal magnetic fields in vivo at 3T because the physiological field amplitudes lie below the method's current sensitivity thresholds.

Original authors: Capiglioni, M., Tabarelli, D., Tambalo, S., Turco, F., Wiest, R., Jovicich, J.

Published 2026-02-18

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Capiglioni, M., Tabarelli, D., Tambalo, S., Turco, F., Wiest, R., Jovicich, J.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The Big Picture: Trying to Hear a Whisper in a Storm

Imagine you are trying to listen to a single person whispering in the middle of a loud, chaotic rock concert.

The Concert: This is your brain. It's always making noise (blood flowing, breathing, heartbeats).
The Whisper: This is the tiny magnetic signal generated by your brain cells (neurons) when they fire.
The Goal: Scientists want to build a super-sensitive microphone (a new type of MRI) that can hear that specific whisper without getting drowned out by the rest of the concert.

For decades, we've used a method called BOLD-fMRI to "see" the brain. But this method is like watching the smoke from a campfire to guess where the fire is. It's indirect; it shows where blood is flowing after the neurons fire, not the firing itself. It's slow and blurry.

This paper asks: Can we build a microphone sensitive enough to hear the neurons firing directly?

The New Tool: Spin-Lock MRI (The "Tuning Fork")

The researchers tested a new technique called Spin-Lock fMRI.

Think of the atoms in your brain like tiny spinning tops.

Standard MRI just watches them spin.
Spin-Lock MRI is like grabbing those spinning tops with a magnetic "tuning fork" and holding them in place for a split second.

The theory is: If a neuron fires right when you grab the top, it creates a tiny magnetic ripple that changes how the top spins. If we can detect that change, we can "see" the neuron firing directly.

The Experiment: The Three-Act Play

The researchers set up a massive test involving 13 healthy volunteers. They wanted to see if this new "microphone" could hear the brain's whisper during a visual task.

Act 1: The Stimulus (The Flashing Light)
They showed the volunteers a flashing checkerboard pattern on a screen. This is like ringing a bell in the concert hall. The brain's visual cortex (the back of the head) should light up and "sing" in response.

Act 2: The Benchmarks (The Control Group)
Before testing the new microphone, they checked two other tools to make sure the brain was actually working:

MEG (Magnetoencephalography): This is a helmet that does hear the magnetic whispers. It confirmed that yes, the brain was definitely "singing" at the right frequency.
Standard BOLD-fMRI: This confirmed that the blood flow (the smoke) was changing as expected.

Act 3: The Test (The Spin-Lock MRI)
Now, they turned on the new Spin-Lock MRI to see if it could hear the same "song" as the MEG helmet.

The Results: A Disappointing Silence

Here is the punchline: The new microphone didn't hear anything.

The MEG helmet clearly heard the brain's song (a magnetic field of about 0.07 nanotesla).
The Standard MRI saw the blood flow changes.
The Spin-Lock MRI? It heard nothing but static.

Even though the brain was definitely active, the Spin-Lock MRI was too deaf to pick up the signal.

The "Why": The Phantom Test

To figure out if the machine was broken or if the signal was just too weak, the researchers built a Phantom (a fake brain made of salt water and a wire coil).

They simulated the exact magnetic "whisper" that the human brain was making.

They turned the volume up on the fake brain until the Spin-Lock MRI finally heard it.
The Discovery: The machine needed the signal to be 3 to 9 times louder than a real human brain signal just to hear it.

The Analogy:
Imagine the human brain is whispering at a volume of 1.
The MEG helmet can hear a whisper at volume 1.
The Spin-Lock MRI needs the whisper to be at volume 3 or 9 just to hear it.
Since the brain only whispers at volume 1, the Spin-Lock MRI is effectively deaf to it.

The Conclusion: Not Broken, Just Not Ready Yet

The study concludes that while the Spin-Lock MRI is a clever idea and works perfectly in a lab (the phantom), it is not sensitive enough yet to hear the tiny magnetic fields of a living human brain at the current strength of 3 Tesla scanners.

What does this mean for the future?

It's a good "No": Science often learns more from what doesn't work. This study sets a hard limit: "We need to make these machines 3 to 9 times more sensitive before we can use them on people."
Hope for the future: Maybe in the future, with stronger magnets or better "microphones," we will finally be able to hear the brain's direct electrical whispers.
Medical use: While it can't hear normal thoughts yet, it might one day be sensitive enough to hear the "screams" of a brain during an epileptic seizure, which are much louder than normal thoughts.

In short: The researchers tried to build a direct line to the brain's electrical activity. They proved the brain was talking, but their new listening device was still too far away to hear it. They have now mapped exactly how much closer they need to get.

1. Problem Statement

Conventional functional MRI (fMRI) relies on the Blood Oxygen Level Dependent (BOLD) signal, which is an indirect, hemodynamic marker of neural activity with a temporal lag of several seconds. A major challenge in neuroimaging is developing methods to directly detect neuroelectrical phenomena (neuronal currents) with high spatial and temporal resolution. While Spin-Lock (SL) fMRI has shown promise in phantom studies for detecting magnetic fields associated with neuronal activity, its in-vivo sensitivity and practicality in healthy human brains remain unproven. Previous studies have yielded inconsistent results, with some detecting group-level effects in visual stimulation but failing to detect signals at the individual level. This study aims to rigorously evaluate whether SL-fMRI can reliably detect and localize human neuronal activation compared to established benchmarks.

2. Methodology

The study employed a three-stage multimodal validation framework involving 13 healthy volunteers and phantom experiments:

Experimental Design:
- Stimulus: 8 Hz flickering quadrant checkerboard visual stimulation (alternating 8s on/8s off) to elicit localized cortical responses in the visual cortex.
- Modalities:
  1. MEG (Magnetoencephalography): Used to directly measure stimulus-locked magnetic fields and estimate physiological field amplitudes.
  2. 3T MRI: Acquired using a Siemens Prisma scanner with a 64-channel head coil.
    - Sequences: Balanced Rotary Excitation (Bal-REX), Balanced Stimulus-Induced Rotary Saturation (Bal-SIRS), and conventional BOLD-fMRI.
    - Parameters: SL frequency set to 16 Hz (on-resonance) and 36 Hz (off-resonance control). Spin-lock duration ( $T_{SL}$ ) was 112 ms.
- Phantom Experiments: An electrical phantom with a loop coil generated oscillating magnetic fields (16 Hz) to determine the minimum detectable field amplitude (sensitivity limit) for the SL sequences.
Data Analysis:
- MEG: Source reconstruction using Dynamic Imaging of Coherent Sources (DICS) and estimation of local magnetic field amplitudes in the occipital cortex.
- MRI Post-processing: A Regression-Filtering-Rectification (RFR) strategy combined with Statistical Variance Mapping (SVarM).
  - Regression: Removed low-frequency drifts and modeled the hemodynamic response to isolate non-hemodynamic variance.
  - Filtering: High-pass filtering to remove residual physiological noise.
  - Rectification: Converted random-phase signals into cyclic signals at the stimulation frequency (0.0625 Hz).
  - Variance Mapping: Used Levene's test to compare signal variance between stimulation and baseline blocks.

3. Key Contributions

Multimodal Benchmarking: This is one of the first studies to directly compare SL-fMRI against both BOLD-fMRI and MEG in the same subjects, providing a ground truth for physiological signal strength.
Quantitative Sensitivity Limits: The study establishes specific detection thresholds for balanced REX and SIRS sequences at 3T, quantifying the gap between current MRI capabilities and physiological reality.
Methodological Validation: It validates the RFR/SVarM pipeline for isolating neuronal signals from hemodynamic noise, confirming that the lack of signal was not due to analysis failure.
Negative Result with Quantification: Rather than simply reporting a failure to detect, the study provides a quantitative explanation: the physiological signal is simply below the current detection limit of the hardware and sequences used.

4. Results

MEG Findings:
- Confirmed robust stimulus-locked activity in the occipital cortex at 8 Hz, 16 Hz, and 24 Hz.
- Estimated the local magnetic field amplitude in the visual cortex to be approximately 0.07 nT.
BOLD-fMRI Findings:
- Confirmed reliable hemodynamic activation in the visual cortex, validating the experimental paradigm and subject compliance.
SL-fMRI (In-vivo) Findings:
- No consistent activation: Neither Bal-REX nor Bal-SIRS produced reliable, stimulus-related activation in the visual cortex after RFR processing.
- While pre-processing maps showed BOLD-related activity, the RFR procedure successfully suppressed these, leaving no significant residual signal attributed to neuronal currents.
- No significant differences in signal variance were observed between stimulation and rest periods for any SL sequence.
Phantom Findings (Sensitivity Limits):
- The analysis pipeline successfully detected oscillating fields in the phantom.
- Detection Thresholds:
  - Bal-REX: ~0.2 nT.
  - Bal-SIRS: ~0.6 nT.
- Comparison: The physiological field amplitude estimated by MEG (0.07 nT) is 3 to 9 times lower than the minimum detectable limits of the SL sequences used.

5. Significance and Conclusions

Current Limitations: Under the present experimental conditions (3T scanner, EPI readout, visual stimulation), spin-lock fMRI does not possess sufficient sensitivity to directly detect physiological neuronal magnetic fields in the human brain. The physiological signals (~0.07 nT) lie below the detection thresholds of the implemented sequences (≥0.2 nT).
Methodological Implications: The study rules out analysis pipeline failure or lack of neural activation as causes for the negative results. The failure is attributed to intrinsic sensitivity limits of the technique at 3T.
Future Directions:
- Hardware/Sequence Optimization: Future work may require ultra-high field strengths, alternative readout schemes (e.g., spiral instead of EPI), or ultra-low field implementations to reduce $B_1$ inhomogeneity effects.
- Target Populations: While healthy physiological signals are currently undetectable, the technique might still be viable for detecting pathological signals (e.g., epileptiform activity) which are significantly stronger and more synchronized.
Benchmarking: This paper provides a critical quantitative boundary for the field, setting a realistic expectation for the sensitivity required to bridge electrophysiology and functional MRI. It serves as a reference for future methodological developments aiming to achieve direct neuronal current imaging.