I have been using the Radial Reslice plugin (as part of Fiji) to analyze CT images of a test object. I noticed a very strange occurrence and I was wondering if anyone could identify the issue. I have inspected the source code of the Radial Reslicing plugin (https://imagej.net/plugins/radial-reslice/download/Radial_Reslice.java) but I am just getting familiar with programming so I can not spot the cause of the issue.
In the first attachment you can see the expected result for my data. In the second result you can see that the result of the Radial Reslicing is tilted exactly 45°.
A few pieces of potentially useful info:
- I have dozens of CT datasets and the problem occurs on about half of them. There is no difference between the datasets that may cause this issue (some of them are identical apart from a small difference in exposure parameters, yet the problem occurs on only of them).
- The problem is 100% reproducible (if it happens on a particular CT dataset, it will always happen, and vice versa).
- The first slice of the radial reslicing is always OK. The problem occurs for all other slices.
- For the Radial Reslicing, I am using a 360 rotation with 0.5 degree increments. These values are required for technical reasons but the problem also occurs for 180 rotation or other degree increments.
- The line which is drawn as input is not drawn manually, it is the result of a previous segmentation process so the Radial Reslicing function is called inside a macro. When I draw a line manually the problem does not occur. However, I compared the entire preceding process, incl. the coordinates of the lines which are drawn, between datasets which work and other which don't, and there does not seem to be any difference...