Fuse Events | StrataQuest Glossary

View

Definition

Fuse events takes two coded images and returns the union of their events as one coded image. Where events overlap, Image 1 wins: its events sit on top and keep their labels, and the events from Image 2 are relabeled starting at the highest Image 1 label + 1. Fuse events (with mask) does the same, then keeps only the result pixels under the white part of a validation mask.

Two variants

A plain union, or a masked one

Image 2 is relabeled

Its labels start above Image 1's highest

Image 1 is on top

Input order determines every overlap

Check for split labels

Overlaps and masks can break events apart

Operations Reference

Operation	Inputs	Description
Fuse events	Image 1 (on top), Image 2	Generates the union of the events in two coded images. Where events overlap, Image 1 events cover Image 2 events. Image 2 events are relabeled starting with the maximum Image 1 label + 1. The result is a coded image.
Fuse events (with mask)	Image 1 (on top), Image 2, Mask	Fuses the two coded images exactly as Fuse events does, then applies the validation mask to the result. Result pixels under the white part of the mask are kept; the rest are set to 0. The reference example uses an 8-bit, 1-channel mask.

Take two transparencies of the same tissue, each with its own outlines drawn on it. Lay Image 1 over Image 2 and look down: where two outlines overlap, you see only the top one. That is the priority rule. A coded image stores one label per pixel, so at an overlap only one event can own the pixel, and Fuse events gives it to Image 1.

Renumbering follows from the same picture. Both detections probably counted from 1, so Image 2 is shifted above Image 1's highest label.

The reference adds a caution. When a large Image 1 event covers the middle of a smaller Image 2 event, the uncovered parts can end up as two or more separate bodies with the same label. The validation mask can cut events apart the same way. The reference recommends a re-label pass afterward, and more strongly after the masked variant; it calls the operation "Re-Label split labels", which matches Re-Label (missing, duplicates)Loading..., the operation that corrects duplicate labels.

Source: StrataQuest 8.1 Engines Reference, ch. 6.4 (pp. 155–162); ch. 6.6 (pp. 194–202).

Simplified

Fuse events combines two sets of detected objects into one. Where two objects overlap, the one from the first input wins. Fuse events (with mask) does the same, then keeps only the part of the result that lies inside a mask.

When to Use Which

Fuse Events (Basic)

Use it to combine two detections of the same kind of object. The reference's example runs nuclear detection twice on one colorectal cancer sample, once with settings for large nuclei and once for small nuclei, then fuses the two with the large nuclei as Image 1. Each detection does what it is good at, and one coded image goes on to measurement. Where both found the same nucleus, the Image 1 version survives, so choose the order on purpose.

Fuse Events (With Mask)

Use it when the fused result should exist only inside a region, for example a tissue area. The mask doesn't determine which input supplies which region. It trims the fused result, both inputs alike, and an event that straddles the mask edge is cut at that edge rather than kept or dropped whole.

Two strategies, two regions

To let one detection cover the tumor and another the stroma, give each input its own region before fusing. One way follows from documented behavior: apply the MaskLoading... operation to each coded image with its own region mask, so pixels outside the region become 0, then fuse the two. If you would rather keep or drop whole events by region instead of cutting them, the Remove objectsLoading... operations select events by overlap with a second image. Region masks themselves usually come from a CompareLoading... or another mask-making step.

After any of these, run Re-Label (missing, duplicates)Loading... so each separate body has its own ID before you measure. In image analysis generally, a measurement table has one row per label, so two bodies sharing a label are measured as one object.

Simplified

Use Fuse events to merge two detections of the same objects, such as large and small nuclei found with different settings. Use Fuse events (with mask) when you want the merged result kept only inside a region; the mask removes everything outside it.

Connected Terms

Connected Component Labeling Category Related
Label Management Category Related
Coded Image Related
Nuclei Detection Related
Morphological Operations Category
Optical Density Conversion Category
Sobel Filter Category
Median Filter Category

Explore StrataQuest

StrataQuest

AI-powered tissue analysis software for spatial phenotyping, cell quantification, and proximity mapping.