Re: Warning of major NHS IT overspend

[Brian Beesley <[EMAIL PROTECTED]>]

On Tuesday 02 November 2004 14:57, Owen Lewis wrote:
> > -----Original Message-----
> > From: [EMAIL PROTECTED]
> > [mailto:[EMAIL PROTECTED] Behalf Of Brian Beesley
> > Sent: 02 November 2004 14:38
> > To: [EMAIL PROTECTED]
> > Subject: Re: Warning of major NHS IT overspend
> >
> >
> >
> > Because it may have advantages in some circumstances - superior
> > resolution,
> > more latitude, ...
>
> The ultimate resolution with emulsion is the granularity of the halide. The
> ultimate granularity for digital techniques is determined by the wavelength
> of the radiation. Think on.

No. You can't build receptors smaller than molecules, even in principle, and 
X-ray wavelengths are shorter than atomic size. 

Practically, CMOS or CCD receptor cells are limited in minimum around twice 
the size of the mask used to etch the chip i.e. 200 nm would be good going.

But, as X-rays can't be (easily) bent to a focus, X-ray imaging is 
essentially shadowing; so the receptor surface has to be the same size as (or 
a bit bigger than) the object being imaged. So you need an image detector 
around 40 cm (400,000,000 nm) across to image a human chest.

No-one can afford a 2 million pixel square detector (4 Tpel). Even 10 million 
pixels in total is unusual and expensive, giving an effective resolution of 
400 mm / sqrt(10,000,000) = 0.126 mm.

Silver halide grain technology is cheap for resolutions to around 0.01mm. In 
fact making grain bigger than 0.01mm is hard, which is why X-ray film plates 
may require more exposure than relatively low-resolution digital sensors.

You can't have it both ways ... with digital sensors you can get more 
resolution - provided you can afford the cost of a large sensor array - or 
more sensitivity - provided you accept relatively poor resolution - but not 
both. As usual, there are physical limitiations.

The manufacturers of digital compact cameras are rapidly running into this 
dilemma. They are now trying to get up to 8 Mpels out of a very small sensor, 
about 1/25th of the area of a 35mm frame; every one of these cameras has 
serious problems with imaging noise, forcing them to reduce the sensitivity 
or use aggressive noise filtering which has other serious effects on the 
image quality. If you really must buy a digital compact I'd strongly suggest 
getting one of the 4-5 Mpel generation whilst they're still on the shelves. 
The alternative approach of using a physically larger sensor has simply not 
occurred to designers of digital compact cameras, since it would mean 
re-engineering the lens assembly, probably making the whole camera bigger and 
"clunkier", and having to have the focussing system work properly (the tiny 
lenses used in most compacts have such a vast depth of field it's almost 
impossible to get wrong!). Bigger sensors are almost certainly inevitable in 
the longer term, though, assuming increased resolution is demanded by 
consumers; however, no one seems to have pointed out that 2-3 Mpel is quite 
adequate for making 6x4 prints, and few consumers ever seem to go for decent 
sized enlargements.

Brian Beesley