So welcome everybody to this session 45 minutes on interventional medical image processing.

No overview today, we just dig into the topic where we stopped yesterday and we, hold on

a second, we looked into the problem of magnetic navigation.

And just a brief reminder for you, we are basically discussing a system that will allow

us to guide the catheter by applying external magnetic forces.

The big picture we have in mind is the neuroradiologist or the cardiologist is no longer required

to adjust the direction of the catheter by complex mechanical motions, but he just uses

the precomputed magnetic field and then you just push in the catheter and it goes the

right direction.

That's the long term vision that we have in mind.

Currently there are systems out in the field where you can use magnetic navigation, but

these systems require some manual interaction.

That means the doctor has to provide the direction of the magnetic field by an interface and

you can stepwise adjust the magnetic orientation of the system.

And that's the system we have built a few years ago while I was still with Siemens.

So you see here, let me just make this full screen, hold on a second, I'll bring it back.

You see here the CM device, this is for instance a standard device in cardiology if you fix

stenosis in the coronary arteries.

And now you go into the control room and here you see the interface that is more or less

operated by a technician and here you see two views from two different directions of

the coronary arteries.

And let me just go back to the two views because that's what we are considering in more detail

right now.

Stop here.

So what you see here are two views of the coronary arteries and the doctor takes one

point here and selects the corresponding point there.

And we all know that if he selects one point here, if we select one point here, the corresponding

point has to lie on a straight line that is defined by the point and the epipolar lines.

And the current implementation requires to select a pair of points in the left image

and the corresponding pair of points in the right image and this defines the 3D orientation

of the magnetic field that has to be generated by the magnets.

So that's what we have to keep in mind.

Now we go back into the interventional suite.

Here you see the magnets, here you see the table and now the magnets come in.

You adjust the direction of the magnetic field and then the system pulls your tip of the

catheter into the proper direction.

Well, that's the system and we are currently considering the difficult problem.

How can I compute the direction of this vector by using 2D views?

So I'm still fighting to get some air and breathing, fighting with my breathing.

So and once again, let me draw the figure of the figures for epipolar geometry.

So the only important idea that you have to know is images are no longer mathematical

structures in the main memory in terms of a matrix.

Images are 2D planes in space that are intersected by the projection rays.

This picture we have discussed yesterday.

So my image is a 2D plane in the 3D space, a 2D plane in the 3D space.

Here is my optical center.

Here is my weld point.

Here is my weld point.

Now it's complicated, but I will start here.

You will see.