ROI Math

The ROI Math command performs a pixel-level boolean set operation between two object classes — an Input class (“A”) and an Other class (“B”) — and writes the result into an Output class. It is used to derive new object shapes from the combination of existing segmented regions, for example to subtract a nucleus mask from a whole-cell mask to isolate cytoplasm, or to restrict spots to the interior of another object.

When multiple Other class objects overlap a single Input class object, they are first unified into one combined “B” mask before the operation is applied.

Parameters

Parameter	Description
Operation	The boolean set operation to apply — see Operations below
Input class	The “A” object set. Each object in this class is processed individually
Other class	The “B” object set. All overlapping objects of this class are unioned before the operation
Other filter classes	Additional classes used to restrict which Other class objects are considered — leave empty to use all Other class objects
Min overlap area	Minimum overlap area (in the chosen Unit) required before a B object is considered to overlap an A object
Output class	Class assigned to the resulting shape
Keep unmatched	When enabled, input objects that have no overlapping B objects are preserved unchanged in the output class

Operations

Operation	Result
AND	Intersection: pixels present in both A and B. If no B object overlaps A, the result is empty regardless of Keep unmatched
OR	Union: pixels present in A or B (or both). The result can extend beyond A’s own bounding box into B’s area
XOR	Symmetric difference: pixels present in exactly one of A and B (the overlap itself is excluded). Like OR, the result can extend beyond A’s bounding box
Subtract	Set difference (A \ B): pixels in A that are not in B. The classic use case is deriving a cytoplasm-only region from a whole-cell mask by subtracting its nucleus

Behaviour Notes

• The operation is applied per A object: each A object is matched against all B objects that overlap it (above the Min overlap area threshold), and their union is treated as the single “B” mask for that pair.
• If Keep unmatched is disabled (default), A objects without any overlapping B are dropped entirely and do not appear in the output.
• The Output class can be the same as Input class, in which case the original object is replaced. If it is a new class, the result is added alongside the original.
• Resulting shapes that extend to the image boundary are flagged as edge-touching, identical to any other boundary object.

Example: Isolate cytoplasm by subtracting the nucleus

Operation:      Subtract
Input class:    cell@whole
Other class:    dapi@nucleus
Output class:   cell@cytoplasm
Keep unmatched: off

This removes the nuclear region from each whole-cell mask. Cells without a detected nucleus are discarded (because no subtraction is possible). The resulting cell@cytoplasm objects contain only the cytoplasmic area, ready for downstream intensity or colocalization measurements.

Example: Restrict spots to within cell boundaries

Operation:      AND
Input class:    cy5@spot
Other class:    cell@whole
Output class:   cy5@incell
Keep unmatched: off

Only the parts of each spot that fall inside a cell boundary are kept. Spots outside any cell are dropped.