Trying to compare specs to determine which system has the best resolution can make your head spin. We recommend getting a demo of the system(s) in which you are interested, providing your own patient or test target, then look at the x-ray to see if you are happy with the quality. Don’t just test the center of the field, check the corners too. If your work requires the best system you can buy, read on.
Example 1: If system A specs 2.5 lp/mm, system B specs 2.5 lp/mm at 20% contrast, and system C specs 2.5 lp/mm at 50% contrast, which has higher resolution? Since A does not specify contrast, we can’t say for sure. It is a reasonable assumption that if system A had a high contrast number it would be mentioned, so we’d eliminate it and go with system C.
Example 2: If system D specs 2.5 lp/mm at 50% contrast and system E specs 3.6 lp/mm at 30% contrast, which is has higher resolution? As lp/mm increase, contrast decreases. These two systems might be the same.
When comparing systems, lp/mm is an acceptable way to specify resolution, lp/mm and contrast (or MTF) is better, and a graph of contrast (or MTF) vs. resolution is the best.
Also be aware that the highest resolution will be in the center of the field. You might want to ask about the resolution in the corners of the field.
Part B, pixels, optics, and resolution.
For our system, you will see our camera has a 9 megapixel CCD, 3056 x 3056. How does that affect resolution? Will 16 megapixels be a lot better? When we speak about line pair per millimeter, it is a very literal specification. If we have a series of black and white bars, and each black bar is ½ mm wide and each white bar is 1/2mm wide, such that the center to center distance of the black bars is 1mm, that is 1 lp/mm. The length of the bars is not important. The pixels on the CCD are aligned in a giant grid in rows and columns, and the very smallest line pairs we can see is when we have one row of pixels being black, the next row is white, the next black, and so on*. For a 17” field** and 3056 pixels this is 3.54 lp/mm. There are not enough pixels to “see” more detail. So seeing 4 lp/mm isn’t going to happen with a 9 megapixel CCD. But what about larger features, say 3.4 lp/mm? If you were to look at a 3.4 lp/mm pattern with a 3056 pixel per line camera, the camera will see it as a 3.54 lp/mm pattern that goes in and out of focus. There still aren’t enough pixels to see that. We need about 1.3 pixels per line in the line pair to be sure what we see is real. Hence any system that shows a 17” field with 3056 pixels is limited to 3.54 lp/mm divided by 1.3 or 2.72 lp/mm. If we switch to a 16 megapixel camera, the resolution will increase by 4/3rds, and we will see a maximum of 3.6 lp/mm. Note this is the limit of the CCD so the numbers in this paragraph assume our optics are good enough to allow this.
If we were to take a 14” x 17” x-ray with our system we will see at best 2.72 lp/mm. What happens if we use software to zoom in on the image and look at the center 6” x 6”? We will see the center 6”x6” bigger, but it still has only a resolution of 2.72 lp/mm. To get higher resolution, we need to change the optics so the 3056 x 3056 pixels are being focused on only the 6” x 6” area. The math above tells us that by doing this we will see 7.72 lp/mm! Go back to part A and recall that contrast should decrease as lp/mm increase. Not if we make an optical change! Our 7.72 lp/mm on the 6” x 6” field will have the same contrast as the 2.72 lp/mm on the 14” X 17” field. This is the beauty of the patent-pending MAGNA-VUE system.
How big of a CCD do we need to get 7.72 lp/mm on a 14”x17” field? The CCD needs to be 8658 x 8658 or 74.98 megapixels. Think about the file size! But would the optics handle that?
Part C, how do the optics and scintillator screen affect resolution?
Simply put, they reduce contrast. With a 12 bit camera there are 4096 levels of grey, 0 through 4095 . In a perfect system, the 3.54 lp/mm described in part B aligns perfectly on the CCD. The CCD should give an output that looks like 0, 4095, 0, 4095, 0, 4095… It is not a perfect world so some of the light, the photons, may not go exactly where they are supposed to. Imagine we are a bit beyond the optics resolution. We might get 3500, 595, 3500, 595, 3500, 595. This is 71% contrast. If we have a smaller pattern, say 1/3 the size, we expect 4095, 4095, 4095, 0, 0, 0, 4095, 4095, 4095, 0, 0, 0, 4095 ….. With these optics we might get 4095, 3500, 595, 0, 595, 3500, 4095, 3500, 595, 0 595, 3500, 4095…. This is 100% contrast. The bigger the feature the more 4095s and 0s we will see. The goal is to make the optics good enough so they are not the limiting factor in the resolution of our system. This allows us to make the best possible use of the CCD. If we make the optics much better than the CCD, it will needlessly increase the cost of the system while providing little or no gain. We have a similar issue with scintillation screens. Cesium Iodide screens have better resolution than GaDox screens but are considerably more expensive. A GaDox screen has sufficient resolution for a 2.72 lp/mm system. However, to boost resolution to 7.72 lp/mm it is necessary to use a Cesium Iodide screen.
Part D, so what is the upshot?
The goal of all systems is to capture as much contrast as possible at the highest resolution and feed the results into the image processing software. The better the data from the imaging portion of the system is, the better the displayed image can be. The intent of this section is to give you, the customer, some information to help you select the best system for you needs. We hope it will enable you to ask better questions of the sales reps. Our goal is to produce a high performing system at a competitive price. If you find a different system that better suits your needs, you should buy it. There are plenty of good systems out there. If you do chose a different vendor, shoot us an e-mail and let us know why. We’ll try and improve so next time we’ll make the sale.
* That is for prefect alignment. Imagine if the bars were perfectly unaligned. Each bar would be ½ on one pixel and 1/2 on the adjacent pixel. The field will appear as totally grey. So a feature at 3.54 lp/mm can be shown absolutely perfect, or totally invisible. This is why the factor of 1.3 is required.
** On a 14” x 17” x-ray, the camera takes an image that is actually 17” x 17”. The top and bottom 1-1/2” are cropped so the system will fit in a standard table.
For information, please call us at 719-282-1587
|
