Sciencemadness Discussion Board

X-ray resolution and sensors

Nixie - 20-12-2007 at 04:54

I understand that resolution is limited by the focal size (cathode projection on anode) but how exactly? For one, the anode is at an angle, so at least in one direction it would seem it would be squashed.
So for a medical rotating anode tube with one focal spot being 0.8 mm, what's the maximum resolution I can expect if used in cone-beam tomography?

~~~~~

Looking at the desktop CT scanners that these guys make:
www.microphotonics.com

There's not much info on what kind of sensors they're using. I need some feedback on pitfalls of my idea to use a scintillation screen imaged by a digital camera (camera being out of the beam path by means of a prism).

-jeffB - 20-12-2007 at 11:52

Quote:
Originally posted by Nixie
I understand that resolution is limited by the focal size (cathode projection on anode) but how exactly? For one, the anode is at an angle, so at least in one direction it would seem it would be squashed.
So for a medical rotating anode tube with one focal spot being 0.8 mm, what's the maximum resolution I can expect if used in cone-beam tomography?


"It depends."

Think about the geometry of your setup, and project rays from opposite edges of your 0.8mm spot through a point in your sample onto your detector. If the point in your sample is 10x closer to the detector than the source, it'll cast a 0.08mm shadow, but you'll be getting a faint image because your source is too far away. If the point in your sample is equidistant from the source and the detector, it'll cast a 0.8mm shadow. If your sample is large relative to the source-detector distance, sample points nearest the source will cast fuzzier shadows, points nearest the detector will cast sharper ones.

In X-ray work, you're effectively working with pinhole optics, and you're always starved for photons. This makes things simpler to analyze than optical-lens systems, but I wouldn't call tomography "simple". I work with people who do micro-CT and other small-scale imaging; if I get a chance, I'll ask after the spot size in our system, which uses a .05mm-pitch (20 pixels/linear mm) sensor and commonly generates images with 80-100 micron resolution (voxel size). I'm not sure what our theoretical limit on resolution would be, but higher resolution implies poorer signal/noise performance, which implies slower image acquisition or higher rad exposure -- it's a multi-faceted tradeoff, and you will weight the factors differently than we do.

Nixie - 20-12-2007 at 12:26

Well, even 0.2 mm is sufficient resolution for me. I'm curious what kind of resolution and sensitivity I can expect from the scintillation screen (one that is pretty standard from what the seller told me). Maybe aiming for a 512^3 is unrealistic (the camera would have to provide at least 9 bits linear and noise free which means combining two exposures per projection if a regular consumer camera is used, and the scintillation screen's grain resolution would have to be small enough). (Note: I'm obviously intending cone beam here; I'll use software freely available or just adapt the one I wrote for optical tomography I did before)

Additionally, how much sensitivity is there of the 3D reconstruction to the monochromaticity (or lack thereof) of the X-ray beam? Filtering (Al plate) helps, but there will still be significant spread in the photon energy I can get out of a system like the one I have, unless I find a way to have fairly flat-topped pulses coming out of the power supply to generate uniform accelerating potential.

[Edited on 20-12-2007 by Nixie]