A note on the json output format

The output format that the docker container requires in order to fully work within the grand-challenge.org environment lead to a bit of confusion, so it’s time to clarify it:

As written in the github repository‘s README, the output format is a JSON file named ‘mitotic-figures.json’ and is expected in the folder /output/ (relative to the root of the docker’s file system). The example given there is:

{
    "type": "Multiple points",
    "points": [
        {
            "point": [
                0.14647372756903898,
                0.1580733550628604,
                0
            ]
        },
        {
            "point": [
                0.11008273935312868,
                0.03707331924495862,
                0
            ]
        }
    ],
    "version": {
        "major": 1,
        "minor": 0
    }
}

Each image is fed independently to the docker container and is thus also processed independently. So the result we see here is for one image only.

In this case, the algorithm has produced two detections:

• One detection with the coordinates x=0.146474 millimeters and y=0.1580734 millimeters
• Another detection with the coordinates x=0.110083 millimeters and y=0.0373319 millimeters.

As a convention, all detections share the z-coordinate of 0 (third coordinate of the list).

The x and y coordinates are measured from the top left on the image, and have been calculated by taking into account the resolution of the image. This is done using the function TransformContinuousIndexToPhysicalPoint() in our reference container.