Near-infrared light penetrates through the bone and tissue the most. This is called the [near-infrared window in the biological tissue](https://en.wikipedia.org/wiki/Near-infrared_window_in_biological_tissue). The reason NIR penetrates in the tissue is because scattering of biological tissue is directly proportional to frequency of the light, hence visible and UV end up scattering a lot so they're unable to reach deep within the tissue. ![[Screenshot 2022-01-11 at 11.52.50 AM.png]] This justifies the use of infrared or above wavelengths. However, as we increase wavelength, light's absorption from water increases (eliminating the use of mid-infrared and higher wavelength light). ![[Screenshot 2022-01-11 at 11.54.19 AM.png]] Hence, an optical window around near infrared region opens up where penetration of light is the maximum. Current light emitters and detectors make it possible to detect backscattered light upto 2-3 cm within the brain (from the surface of the skull). Most NIR is scattered by the tissue but since NIR is absorbed by different biological components differently, we can monitor absorption of the light to deduce the dynamic nature of those biological components inside the body non-invasively. Particularly, NIR is absorbed highly by blood, but importantly it has differential absorption for oxygenated and deoxygenated blood. By measuring how absorption of NIR at two different wavelengths is changing, we can estimate how presence of oxygenated blood is changing at different locations in the brain. And since neural activity consumes oxygen, an increase in oxygenated blood is typically followed by an increase in neural activity (although there's a lag of 5 seconds). This is the basis of fNIRS which estimates brain activity via NIR by monitoring absorption of the light by oxygenated and deoxygenated blood. #### Limitations There are two limitations with this approach: - Temporal resolution: the signal for oxygenated blood flow lags the neural activity signal by a couple of seconds (4-5). Since spiking frequency of neurons can be in kHz range, this lag makes it impossible to pin down exact spiking activity - Spatial resolution: the resolving power of fNIRs is typically in few mm to cm region, which can easily contain >100,000 neurons. Finding out specific spiking activity from this big region is impossible. #### Overcoming temporal resolution via fast optical signals Since 1949, we know that when neurons make a spike, they change in shape (by a few nm) because ions rush in the cell. This change in shape changes the scattering and absorption properties of the neuron. This constitutes what's called a [fast optical signal](https://en.wikipedia.org/wiki/Event-related_optical_signal) of neuron activity because we can detect change in absorption/scattering by the neuron after about 100ms of its activity. In theory and _ex vivo_, fast optical signals have been demonstrated but _in vivo_ and in live animal brains, [detecting this signal has been extremely hard](https://pubmed.ncbi.nlm.nih.gov/15961042/). This is because the intense light scattering within the brain dilutes the signal-to-noise ratio on detector considerably and hence finding a change in scattering due to one neuron's activity is difficult. Other aspects of the living animal such as brain pulsation due to breathing changes the signal intrinsically. #### Overcoming spatial resolution via holograms The core difficulty in detecting single spike due to optical property changes from neuronal activity is light scattering. The backscattered light that reaches detector has travelled everywhere in the biological tissue and hence deducing changes in optical properties of a small region in brain becomes impossible (as those changes could have been due to some other region and light scattered from there to the detector). ![[Screenshot 2022-01-11 at 12.25.28 PM.png]] (via [here](https://www.shimadzu.eu.com/about-nirs-principle-operation-and-how-it-works)) What if we could somehow remove the scattering from the entire path of light emission to detection. This would allow us to focus light on a tiny spot and reliably detect changes in its optical properties by measuring how the detected light changes over time. The company Openwater is developing technology for doing that. The approach they're using is to dynamically create holograms in front of light emitter such that they cancel the scattering (that would have happened if hologram wasn't present). Since they don't know which exact hologram will cancel the scattering properties of the exact brain tissue under examination, they likely cycle through various (randomly generated?) holograms until they find one that maximizes intensity of received light. They focus ultrasound onto a region in the brain and shine the light. The ultrasound pocket changes the color of the light due to an effect similar to doppler shift. Now, a detector for that changed color measures the intensity of light recieved and they cycle the hologram in order to maximize the intensity. This process ensures that the hologram generated optimizes for de-scattering in a way that maximizes received light intensity from the spot illuminated by ultrasound. Now, they can measure optical properties of recieved light in order to find what's happening in the focused spot. In their case, they're measuring blood flow (by measuring absorption of light) but they say they can measure fast optical signals of single neurons (+ also stimulate neurons via ultrasound). Both these claims are in theory and haven't been demonstrated. See their founder's TED talk for details: <iframe width="560" height="315" src="https://www.youtube.com/embed/awADEuv5vWY" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe> <iframe class="signup-iframe" src="https://invertedpassion.com/signup-collector" title="Signup collector"></iframe>