Clarification on Detection Output JSON Structure and Coordinate Reference in Docker Container

I'm implementing a Docker container to output cell detection coordinates, and I'm seeking clarification on a few points about the JSON output format and coordinate reference.

Expected JSON Output Structure: As per the guidelines, my container should output three JSON files:

1. detected-monocytes.json - Contains detected monocyte coordinates.
2. detected-lymphocytes.json - Contains detected lymphocyte coordinates.
3. detected-inflammatory-cells.json - Contains detected inflammatory cells (including monocytes and lymphocytes) as mono-nuclear leukocytes (MNL).

The JSON files should use this format:

{
  "name": "monocytes",
  "type": "Multiple points",
  "points": [
    {
      "name": "Point 1",
      "point": [9.6014, 7.9832, 0.25],
      "probability": 0.92
    },
    {
      "name": "Point 2",
      "point": [13.0106, 11.5523, 0.25],
      "probability": 0.92
    },
    ...
  ],
  "version": { "major": 1, "minor": 0 }
}

For cases where only overall inflammatory cell detection is performed, we should submit empty detected-monocytes.json and detected-lymphocytes.json files.

Main Question: - Coordinate Reference: Should the coordinates in the points array be referenced to each specific ROI, or should they be in the context of the entire WSI? This is important as it impacts how the centroids are calculated and reported.

Thanks for any clarification you can provide!

Hi! They should be relative to the whole slide, with (0,0) being the top left corner.