Okay, welcome. Now we can start.

So, we have the three big chapters

that we are currently considering.

We talked about modalities.

We talked about image preprocessing

and acquisition-specific preprocessing.

And now we are talking about what can I do

if I have multiple images from a single modality?

Can I generate higher dimensional information

and visualize that.

So the reconstruction problem is something

we are looking at now.

And we have discussed the basic important concepts

for 3D reconstruction.

We talked about X-ray based computed tomography.

So how can I use X-ray images to compute

higher dimensional information by solving

integral equations that follow directly

from the X-ray attenuation law.

We have seen the Fourier slice theorem

and we have seen the algebraic reconstruction methods.

And we have seen yesterday how we can transform

the basic equations that we know from

the algebraic reconstruction into a probabilistic framework

and use instead of the least square estimator,

for instance, an approach that makes use

of the Kalbheg-Leibler divergence

or a version that is symmetric

that was the Jeffrey divergence

that we have considered yesterday.

Yeah?

So, and today we want to talk about

a probabilistic reconstruction method

that is applied in systems today.

So if you go to the hospital,

if you go to the clinics of nuclear medicine,

and if you get an image acquired there,

these systems, these huge systems, commercial systems,

they make use of probabilistic reconstruction methods.

So they make use of the methods we are discussing now.

Let's say five to eight years ago,

these systems also have applied

the Fett-Camp reconstruction that we have discussed

with the cone beam reconstruction,

you remember with the projection matrices

and the back projection algorithm.

And it turned out that the artifacts you get in,

especially if you are in the field of nuclear medicine,

molecular imaging, that the artifacts,

they destroy the overall image quality