Okay, good.

So good morning everybody.

Today we have on the program MRI image preprocessing.

So what can we do to improve the image quality in MRI imaging?

We consider this as a very important topic.

The elimination of inhomogeneities.

If you talk to physicists working in the MRI field, they usually don't care about these

things.

Because they tell you, you know, you can manipulate the acquisition protocol in a way that you

can get rid of these things.

From a theoretical point of view, that's right.

From a lab point of view, that's also right.

From a clinical point of view, we have to say, as far as we have experience with MR images,

the correction of MR images is very, very important, especially if you apply post-processing

methods like segmentation, you try to find particular regions in the brain or something

like that.

It's a mandatory, do you say mandatory?

Mandatory.

Mandatory.

It's a mandatory, thank you.

I will have my Starbucks vision.

Otherwise it sounds like mandate.

Okay, mandate.

Mandatory.

Mandate.

Okay.

Mandatory.

Thank you.

And so it's a must and we will learn a lot of methods that can be applied and as usual

in image processing there is not a first choice method.

Usually we have the situation that people work on a new method, a new algorithm, and

then they re-implement the method of all the others and then they compare their method

with the methods of all the others and then they show that their method is superior.

Most of the time the quality of the implementation is approved because for your own idea you

spend two years to implement it, for the software of others you spend let's say half an hour

or so to just have a quick hack and then of course your method is more sophisticated.

We resolve this problem by using libraries like ITK or other computer vision libraries

and then you can compare the methods and it turns out that there is not a first choice

inhomogeneity correction method for all MR images.

It always depends on what your particular problem is and what you are basically required

to solve.

So this semester we talk about very interesting methods that are applied in diagnostic medical

imaging.

We know about modalities, I don't need to tell you anything about that anymore.

We started now a huge chapter on preprocessing.

That's basically a chapter where we transform images into images and we try to improve the

image quality.

So, basically, we tackled the question what can I do with a single image?

In the next chapter that we will start in, let's say, two weeks from now, we will talk

about what can I do if I have multiple images, can I compute higher dimensional information?