Hacker Timesnew | past | comments | ask | show | jobs | submitlogin

I did some work on DBS devices.

Yeah, that's an intense surgery.

The MD doing the implantation was a rockin' bitch-ass motherfucking stone cold killer. Sorry for all the curse words, but that's the best way to describe her. I've never been near someone so dead calm no matter what. Just ice in the veins. She was ~6'5" and wore 6" stilettos all the time. That surgeon had presence. I'd never want to meet her in a back alley late at night.

We were developing a new technique to help out with implantation. One of the hard parts of DBS is knowing where to leave the stimulating end of the electrodes. You want to hit a specific region, but as everyone's brain is different, there's no way to stereotype the process. We had a little optical sensor that we were trying to use to guide the electrodes and see differences in the neuronal density. The place you leave the electrodes is in an area of relatively high density (S. Nigra). Bunch of Fourier analysis, machine learning, and machine vision went into it.



ML for how to direct implantation, or the signals it produced?


I was working on a guiding fiber optic cable, so the signals were live during implantation. The fiber is then pulled out once the electrodes are in the correct position.

We used ML to help determine where the tip of the electrodes were in the brain. The received images needed a lot of processing (Fourier domain mostly), then ML against cadaver brains and other surgeries to help determine the position. Final say included a lot of test shocks to the patient and the surgeon's call.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: