Good morning. Tuesday morning. As usual, we will have first of all an overview of the

lecture to catch up with the current storyline of the lecture and then we will continue where

we stopped yesterday we discussed virtual endoscopy and light fields. So what is

interventional image processing about? Well first of all you should be able to

distinguish between diagnostic and interventional diagnostic and

interventional IP. Question to the audience what is the difference? Right

because you use the machine during the procedure while you are treating the

patient. And in the first week when I was on a business trip we learned a

little bit about segmentation and the structure tensor and the purpose for

that was just to have a way to detect for instance corners. Corners are

important for approaches like we have seen in the chapter on magnetic

navigation. Magnetic navigation. What was the purpose of magnetic navigation?

Anyone in the audience who has an idea what magnetic navigation was good for?

Johannes? So we have magnets and with these magnets and the orientation of the

magnetic field we were able to apply a force on the catheter tip and by this

force you can steer the catheter instead of the mechanical devices that you

usually have while you do a procedure like intervention in neuroendrology or

cardiology. And we were focusing in this chapter on the development of a user

interface for adjusting the direction of the magnetic field. And in this context

we had this very important image. What is it? What is it?

Kerstin, what is it? That's the epipolar geometry and if we look at the image

plane 1 and image plane 2 in 3D we can derive a epipolar constraint out of that

which tells us that P and Q are related by P transposed epipolar

essential matrix times Q has to be 0 and this has to be valid for all the point

correspondences. And based on that we can compute the extrinsic camera

parameters and then we have the points and then we can compute the vector a

3D displacement vector by two projections of this vector in two

different planes. That was the idea. Once a student was asking me the question

what happens or why do we need epipolar geometry? I mean I can do a calibration

of the system for two positions. I rotate with my C arm exactly at this point and

then I know the geometry and I compute the difference vectors of two points

that are clicked and then the 3D structure. The reason for the necessity

of the essential matrix and the epipolar geometry in this context is that for

different patients you need different views and so you need the flexibility

that you can look from arbitrary viewing directions on the patient and that you

can use different projections for the adjustment of the orientation of the

magnetic field. Okay so there was a huge chapter and we described the basic

concept. I mean basically I could teach a whole lecture on epipolar geometry and

related algorithms. So there is much more in it than we have

discussed here. It was just to know a little bit scratching on the surface

what we did here. So if you want to use it in practice you need much more

knowledge than what we have presented here especially how to deal with data

and and how to do the estimation and how you set up all the equations. We

have seen a very basic approach and the drawback of this basic approach is that

in practice it will not work properly. Okay so keep that in mind. If you try it

and you implement it and it turns out that it does not work as I have discussed

it in the lecture do not be worried about that. That's the truth.

Okay then okay we can now navigate with a catheter and adjust the magnetic field

and the next question was well maybe I need another tool that is useful

for interventions, ultrasound. So if I do something in the inner of the