By the next day, we've settled into a kind of pattern. I catch the bus, say hello to Simon Gray and a couple of others at the bus stop. They seem to be 'more than just friends' and are not especially sociable with me.
Simon talks about his time in Brant and that he works in the pharma division. When we get to the Brant complex he even leaves the bus at a different stop. I'm soon at my stop and meet Hermann and Rolf on the way in. They have come along together in Schmiddi's car. We get into the Lab where Juliette is already working on the Cyclone - It looks slightly different to the one that Rolf and Hermann showed me yesterday.
Juliette speaks, "This Cyclone headgear records brain signals without the need for surgery and can either measure the electromagnetic fields generated by groups of neurons or detect small changes in blood oxygenation, which correlate well to nearby neural activity."
This is a different story from the one told by Rolf and Hermann and I'm both intrigued and confused.
Hermann smiles, "I can see you have been paying attention. This different helmet - Call it Cyclone 2 - has a whole different software system and and we are in the process of combining the two systems. Call the hybrid Cyclone 3, but the effects give a much better resolution."
Juliette smiles and adds,"We are using magnetometers to measure tiny changes in magnetic fields and light pulses through the skull and into the bloodstream in order to measure how much oxygen the blood is carrying at any given time. It's a similar concept to the way that smart watches measure blood oxygenation but has a vastly extended coverage. The headgear takes advantage of the relative transparency of the skull and brain tissue to near-infrared light by beaming photons through the skull and measuring their scattering and absorption, allowing inference about blood flow and oxygenation. That's something called haemodynamics."
Hermann adds, "This headgear offers the resolution and sensitivity of state-of-the-art haemodynamic systems across the top layers of cortical tissue.
Hermann produces an animated diagram, "Traditional 'continuous wave' near-infrared spectroscopy devices apply light to the head continuously, which then scatters throughout and is detected at various locations upon exiting the head. It has a decent level of accuracy but the processing time (we'd call it latency) of the system means it is like having very slow reactions. In fact they are more like the reactions from someone on the other end of a satellite phone call."
I was thinking about some of Brant's early uses for the tech. They wanted to deploy it in military scenarios. It would be like using slightly dim low-fidelity mogadon-zombies in a war-zone. Hopeless.
Hermann continues to explain, "Changes in the detected light intensity allow inference of optical-property changes inside the head, like those resulting from neural activity. An analogy would be using a sonar device to detect the movement like shoals of fish in a lake."
I thought this all sounded too slow and only good for tracking big objects.
Hermann puts up a new diagram, "Time-domain systems capture a much richer signal by applying the light in short pulses and precisely capturing the arrival time distribution of scattered photons for each pulse. On average, photons that arrive later travel deeper through the tissue, which reveals additional depth-dependent information about the optical properties of the tissue - allowing for more detailed inference of brain activity."
Inference - there was the word again. It was still mainly guesswork.
Hermann triumphantly concludes, "In other words, this Cyclone headgear is about the most advanced non-invasive haemodynamic and photon inference machine-to-brain communication device anywhere."
I'm thinking it is a solution looking for a problem.
Juliette adds, "With all these challenges, we’re very optimistic that our bionic future. BCIs don’t have to be perfect. The brain is amazingly adaptive and capable of learning to use BCIs in a manner similar to how we learn new skills like driving a car, using varifocal lenses or using a touchscreen interface. Similarly, the brain can learn to interpret new types of sensory information even when it’s delivered non-invasively using, for example, magnetic and light pulses.
I was thinking that it was still a case of join the dots, but someone has erased some of the dots too.
Too slow. Too much guesswork. Over-reliance on predictive capabilities. All before the system was hooked up to the shonky AI system. By itself, I couldn't see this working, but I couldn't tell them so. Not in my first week.
No comments:
Post a Comment