Slashdot Mirror
Facebook Exec's New Startup 'Open Water' Targets Wearable Brain Imaging (xconomy.com)
gthuang88 writes: Display-tech guru Mary Lou Jepsen is leaving her post at Facebook/Oculus to work on a new startup called Open Water. Jepsen, a veteran of Google X and the MIT Media Lab, says the company will develop wearable MRI devices that could help doctors do early detection of cancer and neurodegenerative diseases. Inspired in part by musician Peter Gabriel, Open Water also hopes to use advances in neural imaging and brain-machine interfaces to create a system for reading and communicating human thoughts electronically. She believes there's huge potential in the manufacturing plants in Asia that are primarily used to make OLEDs and LCDs. "My big bet is we can use that manufacturing infrastructure to create the functionality of a $5 million MRI machine in a consumer electronics price-point wearable. And the implications of that are so big." At that price-point, every doctor's office in the world could afford such a device and use it to detect early stages of neurodegenerative disease, cancer, cardiovascular disease, internal bleeding, blood clots, and more.
EM shielding is still essential, and nothing has really developed to change this. The signals are incredibly weak, and extensive RF shielding is required. To give an example, an incandescent light bulb in the scanner room which is reaching end-of-life can produce so much RF from micro-arcs on the failing filament that it can completely swamp the signal.
Capital cost is still very high, typically in the region of $1.5 million for a 1.5T machine, and $2.5-3 million for a 3T machine. Capital costs for the more capable machines are going up due to various developments - e.g. parallel receiver channels (up to 256 channels in the latest machines), parallel transmit channels with higher pulse powers (currently 2x 40 kW RF power amps and transmit antennas, but systems with 4 or 8 transmit channels are in development). Not to mention that there is a push towards larger dimension magnets, which substantially increase the capital cost.
There has also been considerable development in new algorithms for faster imaging, by using incomplete or overlapped imaging acquisitions, which requires extremely complex mathematical processing. CT scanners went through this development a few years ago to improve quality and reduce radiation dose, and image reconstruction went from a 1 ms task running in software, to a 250 ms task running on a 20U rack packed with $250k+ of GPUs. It is likely that the next generation of MRI scanners will use similar compute hardware.
All MRI scanners made in the last 5 years are zero-helium loss, so should not require any top-up of helium. However, they do need heavy chillers to recondense the helium - chillers with cooling powers of up to 1 W are now routinely used, which bring with them energy costs of around $20-30k per year. Many manufacturers also offer helium-free magnets (essentially the magnet coils are bonded to a giant copper/aluminium thermal mass, which is bonded to a cryochiller), these are substantially more expensive in capital cost, and energy cost, and much less tolerant to power failure. However, for areas where helium fills are impractical or too expensive, then these are a viable option.