So welcome to the Tuesday session.

As I pointed out,

we will now stop image preprocessing

and acquisition specific preprocessing.

We could have a whole lecture on that,

or a whole series of lectures on that.

But I think it's now time to switch to a new topic.

And this topic is the third column of our lecture here.

And this is about 3D reconstruction.

So diagnostic medical image processing.

We discussed modalities.

In the second stage, we discussed pre-processing methods.

Basically, to enhance the images, just to remind you,

image undistortion, defect pixel interpolation,

elimination of inhomogeneities in MR images.

And now we talk about 3D reconstruction.

So we consider the problem.

OK, we have now acquired a single image.

It's preprocessed.

Now we have a chance to acquire more than one image.

Is there some additional information

in these multiple images that I can

use to generate other information

for diagnostic purposes?

And the answer is, yes, there is additional information.

For instance, in X-ray images, you observe more or less

integrals.

And you can compute the original function out

of the integrals in terms of a 3D reconstruction method.

And this will become much clearer in a few seconds.

And the fourth part, that's what we

are going to consider in January or part of January and February,

is on image fusion and registration.

And this is the question, OK, now I

have single images, enhanced single images.

I had multiple images.

I computed higher dimensional information.

And now I want to combine images from different modalities

to support the diagnostic procedure.

And this is part of the fourth chapter, key chapter.

And this is on registration and fusion.

So here we have single images.

Here we have multiple images.

And here we have data sets from different modalities.

For instance, modality one and modality two.

And we want to combine these things into a joint coordinate

system such that the doctor, the physician,

can make use of the advantages of the various modalities

at the same time.

That's the idea.