Stage wobble of mesoSPIM V4 setups
The Physik Instrumente M-112K033 stages suffer from high roll tolerances which can lead to wobble artifacts in mesoSPIM V4 datasets.
XZ reslices of datasets exhibit a "wobble" which is visible in features which are elongated in the z-direction.
The spatial period is 500 µm - exactly the pitch of the lead screw.
The thinner the light-sheet (the better the ETL parameters were set), the harder it is to spot this artifact. Usually, badly cleared samples with lots of bubbles (which appear in many z-planes) make it easy to spot.
Longer sample holders amplify the roll motion due to a longer sample arm and lead to larger visible amplitude in XZ-reslices.
As the artifact is due to manufacturing tolerances, each M-112K033 stage has a different roll amplitude.
The Physik Instrumente M-112 stages are extremely common in light-sheet microscopy (going back to the time of the first SPIM generations in Ernst Stelzers lab at EMBL) and were used in the Clarity-Optimized Light-sheet Microscope (COLM) by the Deisseroth lab. To accomodate large cleared samples, PI was asked to extend the travel range from 22 mm to 44.5 mm which lead to the custom M-112K033 stage. As PI stages were performing very well in the COLM setup in Geneva, we decided to use them in the mesoSPIM V4 design as well to ensure compatibility between the setups.
After setting up the first 5 mesoSPIM V4, we noted that 2/5 setups exhibited annoyingly high wobble artifacts which is why the mesoSPIM V5 was designed with better PI stages.
The root cause of the issue is that the linear bushings in the stage are unsuitable for such long travel ranges. Therefore, the mesoSPIM V5 was designed with stages based on crossed-roller bearings.
Minimizing the impact on tiled aquisitions
As the stage roll affects both the X and the Z-axis, this problem is apparent during lateral movements as well: During X-movements, the sample appears to undergo "breathing" - slight periodic z-movements that are very apparent when the ETL parameters are set properly. To simplify stitching of large datasets, it is recommended to use integer multiples of 500 µm as lateral offset For example, tiling at 4x using the MVPLAPO1x objective is usually done using 2500 µm offset (at 3.29 mm FOV in CLARITY-cleared samples).
As the Y-stage (PI L-406) is not affected by this issue (different bearing types), there should be no apparent breahing during Y-movements (vertical sample movements).
Use the wobble correction code developed by Rob Campbell in the mesoSPIMmat toolbox to do post-hoc correction of datasets.
1. Exchange X and Z-stages
The X and Z-stages are identical. Select the better stage for the z-axis by testing them both as a z-stage.
2. Upgrade the stages
The mesoSPIM V5 is using PI L-509 stages which minimize this issue.