timz
New Member
Posts: 14
|
Post by timz on Oct 1, 2013 12:05:09 GMT -5
For those who have had experience with the system, can you comment on what is the top priority to improve, or to change if given a chance? I think this will be helpful for others who have not build the system, so that they can avoid the pitfalls.
|
|
|
Post by JDazai on Oct 1, 2013 14:20:43 GMT -5
Hi Tim,
If we were to do this again, I think we would incorporate a manual rotation stage under the focus translation stage (the one parallel with the optical axis) and also incorporate a manual tilt stage under the z-stage. Although these are not necessary, both of these would make the calibration step much easier.
Since building the initial setup outlined in our paper, we have added the following:
- Shuttered light source to minimize photobleaching. - Mounted the fluorescent light source to the focusing stage to move with the specimen when being focused. - Implemented a interchangeable focusing lens for the fluorescent light source for various sized specimens.
These are just a few examples of things that can be implemented once the initial base system is constructed.
Jun
|
|
|
Post by JDazai on Oct 1, 2013 15:45:25 GMT -5
I would also like to add that a 2' x 5' breadboard is probably unnecessary and that you could get away with a 2' x 4' one if you were to only build the system outlined in the paper. However, if you wish to have the ability to modify it in the future, having a larger breadboard may be beneficial.
|
|
|
Post by mwong on Oct 3, 2013 12:43:38 GMT -5
On the software front: If one was to try to optimize the software, one can look into the following things - At the time of building the Labview software, we were not able to find a way to export 16-bit TIFF files for each projection image. Therefore, we ended up exporting .bin binary files of the 12-bit raw data captured by the CCD. Because of this, we wrote a Matlab script to automatically convert the .bin files to TIFF files. As well, image averaging and darkfield subtraction is done in the Matlab script as well. It would be of great benefit to perform all these tasks within Labview and to export 16-bit TIFFs such that they can be sent to reconstruction (NRECON) without matlab as the middle man. Of course the extra processing within Labview would have to be done is such as way that it did not lengthen scan time.
- In the Labview GUI there is a separate Live View and Scan View window that displays images while pre-scanning and scanning respectively. This, at least to our knowledge, needed to be separated for our algorithm to work. If someone could find away to use only one window for both that would be beneficial.
- The calibration algorithm is built in Matlab such that a scan of a ball bearing needs to be taken through Labview and then analyzed in Matlab. Again, if this could be incorporated into the main Labview GUI that would get us closer to our goal of a standalone software for this OPT
|
|