The phenomenon of fake resolution
June 05, 2023 | Andreas Deresch, Gina Naujokat
By chance, we recently encountered a phenomenon together with a customer that we had not paid attention to before. An apparently resolved X-ray image turned out to be a fake. A simulation model proved the mystery.
Our customer showed us an X-ray image of a JIMA resolution structure at 0.4 µm line width.
JIMA resolution structures are among the best-known image quality indicators (IQIs) for checking the resolution of image chains including microfocus X-ray sources in use. The smallest segment of 0.4 µm is used in corresponding microfocus applications with focal spots of less than 0.7 µm full peak width at half-height*. The structure consists of six lines of 0.4 µm each**, seven spaces of 0.4 µm and a wider frame.
The apparently well-resolved X-ray image perplexed us with its unexpectedly blurred frame of the structure.
This prompted us to do some experiments. For comparison, we took scans of three different structures of the corresponding JIMA test specimen and created a profile of the marked area in each case.
At 0.7 µm, we obtain a resolved image with the chosen focal spot in which the line structure is clearly visible, and also the profile shows six lines.
At 0.6 µm, all six lines are also imaged, although the noise has already clearly increased.
Even at 0.4 µm, we obtain a resolved image by subsequent image processing (increasing the contrast, which, however, also increases the noise), showing all lines clearly in the profile, too.
But our input image shows only five instead of six lines in the profile. It must be considered unresolved because the image information merges, and a distorted structure is depicted.
Using simulation software, we recreate this phenomenon with a simulated, uniformly bright, circular focal spot of 1.7 µm in diameter and obtain the same result.
We are now interested in how different focal spot sizes, each given here as the diameter of a uniformly bright circular disk, behave for the 0.4 µm structure and focus on the number of lines identified. In 0.1-µm steps, everything runs smoothly up to 1.0 µm.
0.1 µm focal spot – 6 lines:
0.6 µm focal spot – 6 lines:
0.8 µm focal spot – 6 lines:
1.0 µm focal spot – 6 lines:
1.1 µm focal spot – NO SINGLE LINE VISIBLE!
1.2 µm focal spot – ONLY 5 LINES VISIBLE!
Remaining 5 lines up to the focal spot of 1.7 µm until a weak visibility at 1.8 µm:
At 1.9 µm focal spot, again, no lines can be seen anymore:
At 2.0 µm only 4 (weak) lines appear,
which are then very clear at 2.1 µm:
We were able to reproduce this process until only three lines were visible at 2.8 µm,
which then became clearer again. As a reminder: The JIMA structure contains 6 lines! All other representations do not correspond to reality.
This phenomenon therefore seems to be obviously dependent on the shape / size of the focal spot used and could be numerically reproduced on the basis of the simulation model.
Looking at the modulation transfer function (MTF) for an image with the simulated focal spot, one can see several spatial frequencies, corresponding to different line widths of the resolution patterns, at which the curve reaches the zero line, i.e. the image would show no contrast at all. The areas between two such points correspond to the visibility of a reduced number of lines. Only in the range up to the first(!) falling below a selected threshold value the corresponding structure size can be considered as resolved. A common threshold value oriented to human perception is 10%.
With this simulation we wanted to illustrate that apparently resolved images of a test specimen do not necessarily always show what is present. In the case of using the different JIMA test specimens, it is highly recommended to verify the actual presence of all lines of the structure. Comet Yxlon offers in its CT systems an operator support software integrated in Geminy, which prevents misinterpretation and always ensures what resolutions can actually be achieved! You can see an example of a resolution test report below. Just contact us.
* The determination of focal spot sizes in this size range has not yet been standardized. As part of the NanoXSpot funding project, two methods were further developed with the participation of Comet Yxlon and proposed for standardization.
** Nominal. There is a tolerance of ±10% at 23°C. An accompanying certificate contains the actual sizes measured at final acceptance.
September 22, 2023 | Gina Naujokat
View three simple (and fast!) methods for evaluating porosity via CT with Dragonfly. Each method can be automated using macros, with a downloadable application note detailing the method and analysis time required.View more
July 27, 2023 | Gina Naujokat
A sedimentary slab from the Holstein Rock was to be examined for its fossil content using computed tomography. Latest CT technology provided results no one had expected.View more