Wednesday, February 04, 2009

Dark Helmet

First up, Sir Terry Pratchett is a legend, and I've read all his books, including the ones nominally for children. The programme was both moving and watchable. He did an excellent piece for, IIRC, BBC Breakfast when he said that he wasn't sure it works, and discussed the importance of controlled testing. "It" is a therapeutic light helmet that, it is suggested, is beneficial for Alzheimer's sufferers. 

The bloke pushing the helmet came across as a bit creepy, to be honest. I'm going to remain skeptical about the idea for the moment: there's been one study in mice (here, thanks to science based medicine), and an unpublished, non-controlled human study which formed the basis for a press release.  Hmmmm.

As I'm not qualified to discuss the medical evidence, such as it is, I'll confine myself to the photonics [1]. I was a bit surprised by two aspects of the helmet: (i) the choice of wavelength and (ii) the sheer weight of thing.

(i) 1072nm (the wavelength from the mice study) is a bit of a funny wavelength to chose to generate with diodes.  You could could get a nice neodymium:YAG laser (suitably attenuated), diode pumped straight from the mains at 1064nm (and maybe externally tune to push the wavelength out to the edge of the gain spectrum), I suppose, but diode-wise the wavelengths 1050nm-1260nm are in a bit of a wasteland. Plenty of diodes in the red-NIR (the AlGaAs/GaAs  heterostructure laser was born to serve this range, and bulk GaAs has a band gap ~850nm at room temperature), you can hit the green-blue and even UV by making the jump from the zincblende to wurzite structure - with corresponding leap in bandgap - by putting some nitrogen into your device, and the telecoms wavelength region of 1.3micron (dispersion minimum of fibre) or 1.55micron (attenuation minimum of fibre) are well served (and correspondingly cheap), but the range in between? InAs/GaAs quantum wells, I suppose, might do it, or more exotically a quantum dot laser that either operates in an excited state or has been annealed to move the ground state to the appropriate energy; or perhaps an InP heterostructure? Could be pricey. I note the company already produces a hand held cold sore device that operates at this wavelength, so presumably they have access to a source of these diodes. The company sells it's device for £45 - I wonder how many diodes it contains? 

(ii) for goodness sake, the thing left impressions on his skull, and had a least one bloody great cooling fan. It would be much more sensible to build a fibre-based system: generate the light on a box on the floor, generate all the waste heat there where it can be air cooled away. This would make wearing the helmet more comfortable, and make controlling experiments easier (you wouldn't know if the devise were switched on or not), and would allow a range of wavelengths to be tested. This would be important to find out what is going on, and if a cheaper kind of light source could be used. I know fibre isn't brilliant away from 1.3 and 1.55micron, but it only needs to travel a few meters.

Bottom line: we need more information, some double-blinded controlled experiments that can survive peer review - and a better helmet design.

[1] If you are a UK taxpayer, perhaps I could take this opportunity to thank you for funding me to study this stuff. I'm very grateful. 

No comments: