The following content has been provided by the University of Erlangen-Nürnberg.

Okay, so welcome everybody.

So let's see if we get an image.

Can you see the slides already? Okay, so welcome everybody.

Welcome to Interventional Medical Image Processing.

And today we will have the last, not the last lecture, but the last content lecture of this term.

And we will only have a final lecture, a last lecture on July 12 at 4 o'clock.

And here we will have the topics of evaluation and questions and answers.

And before you leave today, I have the transaction numbers here.

And everybody can pick a transaction number when you're leaving.

So I'll be handing them out here, and everybody gets one.

Good. We were talking about ECG-gated reconstruction and motion-compensated reconstruction.

So last lecture we learned that all of the reconstruction theory that we have is essentially assuming that we have a static object.

Now, unfortunately, patients are not static objects, and we have to do something about it.

And one idea that we looked at last time from 2006 is that you virtually freeze the heart.

So we can do a reconstruction by virtually freezing the heart.

And we assume that we have the same heart motion in every heartbeat.

So in order to reconstruct, we just take the heart, but from different projection angles.

And we only use the projection angles that were acquired in the same heart phase.

Then we put them all together, and we can reconstruct.

And the reason why we were doing this is because typically in the interventional scenarios, you have scanners like the one shown on the slide.

And this one is rotating around the patient.

And of course, it has an open gantry.

So you can see the detector and the tube mounted to the system.

And you cannot rotate as quickly as you would do in a diagnostic scanner.

In a diagnostic scanner, you can rotate about the patient four times a second.

So there you can even take an image of the beating heart, a volumetric image of the beating heart, in let's say 60 milliseconds.

However, in a system like this one, you have to rotate at least 200 degrees.

And you cannot do this in a fraction of a second.

And the reason why you can't do it is mostly because you don't want to hurt anybody.

So the system could move technically really fast.

It could do the motion maybe in one second or in two seconds.

But typically, you take much longer. And for a high resolution scan, you take up to 20 seconds.

And in 20 seconds, a lot of motion can happen, in particular, if you have something like up to 180 beats per minute of a heart that is beating.

So there is definitely motion.

And now the idea that Laurich and Farik came up with is that you create multiple sweeps.

You rotate the C arm back and forth.

And by doing so, you acquire more and more data of the same heart phase.

And at some point, you cover the entire angular range.

And you can select these projections to do a reconstruction.

And this is actually possible. So you can actually buy this as a clinical product.

However, typically, the heart is affected by a kind of disease.

And if the heart is affected by disease, there is a certain likelihood that it will not be perfectly periodically beating.

So this is the reason why people came up with additional ideas how to compensate for this problem.

And one solution is image-based gating.

And image-based gating is a very interesting approach.

And we can also write this up into a very nice optimization problem.

So let's have a look. Can you see the synograms?

So on the left-hand side, you see a synogram of an object that is static.

And you can really nicely see the sinusoidal curves.