Q: How is SOPi pronounced?

SOPi is pronounced as /sōpī/ i.e. 'So' + 'π'.

Q2: What does SOPi stand for and why a new acronym?

SOPi stands for Scanned Oblique Plane Illumination. This is the main imaging approach in our microscope. We coined a new acronym because our technique was distinct enough to have an easy to call name, and we scientists love to name things ;)

Q3: Are there any existing approaches which implemented oblique plane imaging before SOPi?

Yes! There are multiple approaches reported before SOPi which utilize oblique plane imaging. We have learnt a lot from all of them. We highly acknowledge works involving OPM (Oblique plane microscopy), SCAPE (Swept confocally-aligned planar excitation) microscopy, and OS-2P-LSFM (Oblique scanning 2-photon light-sheet fluorescence microscopy).

Q4: That is a big list of prior work! How does SOPi differ from them?

It is difficult to answer that briefly but the methods differ from each other either in their scanner arrangement or in scanning geometry. OPM uses piezo mounted objective for remote scanning of oblique light-sheet. Here the oblique light-sheet scans along axial direction. SCAPE uses a polygon mirror for attaining lateral scanning of oblique light-sheet. OS-2P-LSFM uses a high index glass window for lateral scanning of oblique light-sheet and SOPi uses a planar mirror for lateral scanning of oblique light-sheet. Following simplified animation would help differentiate between OPM, SCAPE and SOPi scanning arrangements.

Q5: I read SCAPE paper and I fail to see the difference between SCAPE and SOPi. A planar mirror based version of SCAPE was already suggested in the original paper.

That is an excellent question and to answer this let's dive deeper. SCAPE approach was developed on similar lines of confocal theta microscopy which utilizes a polygon scan mirror. While that architecture works well for confocal scanning, it has one drawback in low NA illumination and whole oblique section imaging. The light-sheet tilt varies as it scans through the sample. This varying tilt oblique light-sheet makes the accurate 3D reconstruction of scanned volume computationally heavy. Even the planar scan mirror based arrangement of SCAPE was suggested as a version which would increase the system NA allowing for better collection of fluorescence signal. We have not come across any SCAPE implementation trying to solve the scan dependent tilt of the oblique plane light-sheet and thus we took this as a challenge and have performed the scanning architecture optimization in same direction to come up with SOPi. SOPi attains a constant tilt scanning of oblique light-sheet.

The following animation would help better understand the SOPi scanning geometry:

Q6: Why do you make a big deal out of 2P SOPi? 2P SCAPE and OS-2P-LSFM already do that.

2P SCAPE implementation is based on scanning and de-scanning of laser line along two axis to get imaged on central row of a camera. This asks for a post processing step to get the 2D or 3D reconstruction. SOPi utilized scanning and de-scanning along one direction leading to live visualization of 2D sections and the need for post processing arises only when 3D volume reconstruction is desired. While a powerful computer can perform live 3D visualization in either approaches, SOPi can do live 2D visualization with a minimalist computer keeping the cost of whole system low.

OS-2P-LSFM is a nice approach but its design cuts the system NA to much lower values than OPM/SCAPE/SOPi designs thus not allowing for good axial resolution. Moreover, the constant tilt scanning would fail here for moderately longer scan range due to dispersion characteristics of the glass plate.

Q. How does the SOPi acquired data look like? May I get some sample data?

A. Although our publications provide visualization of SOPi data, we understand that playing with real data is a much enriching experience. So we have made some data freely available at https://northwestern.box.com/s/haj6anlqn1f4dzrcotetx453q0368c7p.

A larger dataset is accessible at https://doi.org/10.5281/zenodo.5088088

This data is from our Optics Letters paper openly accessible at doi: 10.1364/OL.44.001706.

Q. Thanks for the data but how to I visualize it in 3D?

A. The image sequence (data) you have represents a 3D volume of the scanned region in the sample. You must know few parameters regarding the data in order to be able to proceed with the steps for visualization.

Required parameters:

1. Scan range (250 µm for given data - mentioned within filenames)

2. Effective sample length covered by 1 camera pixel (1 µm for given data)

3. Light-sheet tilt angle (45° for given data)

Warning: If you have your own data with a different tilt angle then don't apply the provided affine transformation matrix as the file is specific for 45° tilt light-sheet. Read our papers and make your new affine transformation matrix.

You may wish to note down the camera exposure time, frame rate, laser power etc. for having a complete info but they are not required for 3D reconstruction and visualization of the data.

Fiji plugins: There are few important plugins which must be added before proceeding to the visualization steps

1. transformJ: This plugin is useful for the affine transformation of the data. Installed with ImageScience plugin https://imagej.net/ImageScience

2. ClearVolume: This is for 3D visualization of the data. Installation instructions here https://imagej.net/ClearVolume

3. 3Dscript: An intuitive script based 3D visualization tool. Installation instruction here: https://bene51.github.io/3Dscript/

Visualization steps (with Fiji)

1. Trim: If your image sequence contains excess of single sweep through the desired scan region. Find the edges (abrupt jumps) on both ends in your image stack (data) and use *Image>Duplicate* option to enter the frame numbers manually to get a new stack corresponding to single sweep of the desirable region. This step is not required for the provided dataset.

2. Re-order hyperstack: The trimmed image sequence is a time sequence. Convert it to space sequence by Image>Hyperstacks>Re-order Hyperstack. [Use dropdown menu to swap z->t and t->z]. This step is not required for the provided dataset.

3. Image properties: You need to tell the pixel/voxel dimensions here. Set the pixel width and height as per the SOPi system specs on a given day (it is usually 0.5 µm or so). Calculate the ratio of the physical scan range (you need to know this from your SOPi scan session ideally written in your data file name) and number of frames (corresponding to the known physical scan range) and enter this as your sequence voxel depth value. This step is not required for the provided dataset.

4. Scale: This is an optional step but recommended to keep your PC happy. Check the file size of the new stack. If it is large to be handled by your PC (e.g. >500 MB) then scale it along x and y direction by 0.5 (or some other factor as per your case). Find this scale option at *Image>Scale*. Enter X scale and Y scale values (keep them equal). Save the image stack as .tiff on the local drive. This is recommended as next step is RAM intensives and PC is likely to hang for a large dataset [you don't want to lose your manual labor involved with previous steps].

5. Affine transform: Close all the image sequences other than trimmed and scaled version. Use the Matrix file browse button at Plugins>TransformJ>TransformJ Affine to upload the shearNscaleObliq file. The file is available in the same link shared above. Check all three boxes and click OK. [Warning: Resample isotropically option will increase the file size a lot and might be skipped safely in most cases.] Once done with the affine transformation you will get another [larger] image sequence. Save it as a new .tiff file.

6. Visualize: Use either ClearVolume: Plugins>ClearVolume>Open in ClearVolume or 3Dscript: Plugins>3Dscript>Interactive Animation to visualize the affined image stack. If file is too large to load on either, use BigDataViewer: Plugins>BigDataViewer>Open Current Image instead.