Remove bed from CT images

I wonder if there is already a plugin to remove the bed from CT images, otherwise I would like to make one. Can anyone have some advice, algorithms, suggestions, etc on this


Hi @BishopWolf,

I don’t know a FIJI plugin, but I know that commercial software (especially in the field of Radiotherapy planning) have algorithms for that.

A common ansatz is to make a CT image of the bed without a patient (in high resolution) and keep it as reference. Afterwards, this image is rescaled to the resolution of the dataset with patient on the bed and afterwards subtracted. Usually, the coordinate system of CT scanners is fix (meaning, the cartesic origin is a fixed point in space). However, depending on the scanner, the bed may or may not bend if a patient is on it. Thus, applying a registration algorithm may make sense. But I would first try without.


I am currently using a commercial system that does not require a reference image to remove the bed, idk how they do it but I think it is a matter of pattern recognition, since every bed has an U shape. I agree with you a reference image is the quickest answer, lets give it a try!

I found out that averaging all slices and finding stddev for each pixel I can discriminate wich pixels remain unchanged (the bed and the air), and taking only these pixels I can, in theory, segment the bed. Wish me luck! (No, it doesnt worked)

I found this article: Automated patient couch removal algorithm on CT images describing the algorithm, but I have no access to it. Can somebody please send the pdf to me