Field of View

TissueFAXS scans a region one FOV at a time, moving the stage between frames. Neighboring FOVs overlap so that stitchingLoading... can correct stage errors: its BF/FL Image matching method finds the pixel shift between each pair of neighbors. TissueFAXS recommends an overlap of at least 15%, set as Width/Height Overlap in percent or in pixels. The FOV's physical size comes from Calibrate Field of View, done for each objective.

StrataQuest keeps that structure. It doesn't copy the images; it shows and analyzes the original virtual slideLoading.... Running an engine usually means analyzing each FOV of the sample or ROI, typically in parallel and in no fixed order, and image results are stored per FOV: 10 FOVs and 2 image outputs give 2 sets of 10 images. Some engines, such as Tissue DetectionLoading..., work from an image in the cache instead, and their outputs are projected back onto the FOVs.

Two mechanisms deal with FOV edges. Border Input, an option of the Basic Operations Module and of detection engines such as NucleiLoading..., adds a border of pixels from the neighboring FOVs, Border Size pixels wide, so a filter or detector works on continuous data. Object stitching is a step after Detection in every layer set to stitch objects, which is the default. It matches the local labels along FOV borders into global labels, so a nucleus cut by a border becomes one event in the coded imageLoading... data.

Source: StrataQuest 8.1 User Manual §11.1 (pp. 149–151), §9.6 (p. 144), p. 18; Engines Reference ch. 6 (pp. 19–21), ch. 14–15 (pp. 405–410), ch. 26 (p. 452), ch. 33 (p. 480), ch. 41 (p. 538), ch. 44 (p. 550); TissueFAXS 8.1 User Manual §5.1.3.2 (pp. 174–176), §6.2.1 (pp. 209–215).

Sampling. In general, each pixel is a sample of the continuous image the optics form. The NyquistLoading... criterion says the sampling interval must be at most half the smallest distance you want to resolve. With a resolution of 500 nm, a pixel sizeLoading... of 250 nm or smaller is needed, and finer features need finer pixels and a higher numerical aperture. In StrataQuest, Pixel Size is a sample property you can edit.

Shading within a FOV. In general, illumination is not perfectly even across a frame: it often falls off toward the edges (vignettingLoading... is one cause), so identical cells can look dimmer near the edge. Because every FOV carries the same pattern, uncorrected tiles can show a faint grid when placed side by side. TissueFAXS corrects this with Illumination Correction (brightfield) or FL Shading Correction (fluorescence), and StrataQuest has its own Shading Correction, all built from a correction image. Background RemovalLoading... is a different tool, which subtracts an estimated background level.

Why borders are hard. A filter computes each output pixel from a neighborhood, and at a FOV edge part of that neighborhood lies in the next FOV. Border Input fetches it. Where no neighbor exists, at the edge of the scan, the Border Option sets what fills the border: Fill with black, Fill with white, Mirror or Copy last pixels. Detection meets the same problem at a larger scale: a nucleus cut in two is two partial objects until object stitching joins them.

Suppose a TissueFAXS scan of a section gives a region of several hundred FOVs, some of them empty glass along the tissue edge.

1. In TissueFAXS, the overlap is set to at least 15% and stitching is enabled, so the region displays without visible seams.
2. In StrataQuest, Flag for Exclusion removes the empty FOVs from the analysis without deleting them, and the sample's properties show the Excluded FOVs Count.
3. Nuclei detection runs FOV by FOV with Use Border Input on, so nuclei at FOV edges are segmented from continuous data.
4. Object stitching joins nuclei that cross FOV borders, and Standard MeasurementsLoading... (global) measures each stitched nucleus once.

Several StrataQuest engines come in versions named after the FOV structure.

• Local versions work within each FOV. Standard Measurements (local) measures local events, without stitching information, and needs Stitch objects disabled in the layer. Distance MapLoading... Outside (local) stores 8-bit distances and is not suited to areas of interest one FOV apart.
• Global versions use information from all FOVs or from stitched events. Standard Measurements (global) measures global events and needs Stitch objects enabled, which it is by default. Otsu Threshold (global)Loading... computes one threshold from all FOVs. Distance Map Outside (global) stores 32-bit distances and works across FOVs, one FOV per iteration, up to Maximum iterations (default 7).
• Border options let BOM and detection engines add pixels from neighboring FOVs as a border, preserving continuity at the edges.

So the choice is about scope. Use a global engine when objects or distances cross FOV borders. Turn stitching off only when FOVs are unrelated, as with images brought in through Import Folder, the case the User Manual names.