A description of data types for use with caMicroscope
All slides should be openslide compatible, but not all formats seem to work consistently. SVS and TIF file extensions seem to work most reliably.
The prediction and segmentation applications can use a suitable tensorflow.js formatted model and weights to operate. This guide may be useful for the generation of these files.
Segmentations, Heatmaps, annotations, and Templates are all primarily mongo documents, and are described in the next section. Use mongoexport or mongoimport in a docker exec -it ca-mongo context.
Slides are identified by their uuid, which is automatically generated. Slides have a display name name, a file location in context location, a measurement scale conversion factor in micrometers per pixel mpp, and study and specimen fields.
Marks are extended geoJson formatted items with viewport coordinates. The geojson component is contained within geometries. provenance describes the source provenance.analysis.source and associated slide uuid provenance.image.slide of the mark, as well as the mark's execution id provenance.analysis.execution_id, and the mark's display name provenance.analysis.name. properties has any filled out information associated with an active template.
Heatmaps are data on a grid. data is a list of grid elements, in the format [x position, y position, variable 1 value, *any additional values, if any*]. The associated fields are stored provenance.analysis.fields in a list of items of format {"name":"FIELDNAME", "range":[min,max]}. The x and y positions are translated according to provenance.analysis.coordinateSystem. The associated slide, as with marks, is provenance.image.slide. The heatmap relative id is provenance.image.execution_id.
See https://github.com/camicroscope/Distro/blob/master/config/test_seed.js