3D segmentation over time

Tags: #<Tag:0x00007fd541ce60a8> #<Tag:0x00007fd541ce5ef0> #<Tag:0x00007fd541ce5d38>



Does someone know if there is a convenient workflow in ImageJ to segment objects in 3D over time?

We tried the 3D Objects Counter and the 3D Image Suite but neither of them seems to understand hyperstacks as an input (unless we did something wrong), but both only process the first time point.

Of course we could script both tools (or probably even better the imagej-ops), but we were wondering if there exists already something that understands hyperstacks and can produce outputs for quality control, such as segmentation boundary overlays.

(One output we would for instance need from the objects are the Surface/Volume ratios.)


I nowadays use KNIME Image Processing for everything that’s concerning 3D segmentation. It fully supports n-dimensional datasets due to its architecture on top of ImgLib2.

It has an Interactive Segmentation View that’s quite useful, and will hopefully be even more so with the integration of BigDataViewer/ThreeDViewer/ClearVolume in the future.