Welcome back to Beyond the Patterns.

Today I have the great pleasure to introduce Song Weizhou.

So he is currently postdoc at Johns Hopkins University, graduated his PhD in 2021 from

Arizona State University and is very, very active in the medical AI community.

You can see that he has published more than 30 peer-reviewed papers in top venues like

IEEE, TMI, has received several outstanding research awards like the Met.I.A. Best Paper

Award in 2020-10, in 2019 the Mikai Young Scientist Award.

So he is very, very, very active in the field and also he was named, by the way, as a top

2% researcher by Stanford University in the years 2022 to 2024.

So it's really a pleasure to have him here for an invited presentation.

It's entitled Body Maps Towards 3D Atlas of the Human Body and you will see that he is

working really in a big network of researchers that are providing large databases for training

medical AIs and I'm very, very happy to have him here.

So Song Wei, the stage is yours.

Thank you very much for the kind introduction and it's a really honor to be here and to

share our recent work.

I'm currently a research scientist at Johns Hopkins University.

So recently we have an initiative in J-H-U.

It's a data science and AI trusted data set.

The aim is to provide the leading source of database for research and the development

of trustworthy AI systems.

And our research group is creating large-scale CT database by having medical professional

and engineering experts working together to annotate human anatomy and detect the disease

in the CT scan.

And make sure this data set can be public available for the research purpose because

you know there's a whole bunch of reason the medical data is very difficult to share across

the world.

So why annotating human body is so important?

Let's take a look at how we understand the earth.

Human has been creating maps of the earth thousands of years.

And satellite images again being used from 1960.

This is a satellite image of Johns Hopkins CS department building.

But you see there's not many useful information from the raw image data because you are before

you annotate anything.

And what Google did is they annotated the building, the grass, the trees, the mountain

and to start to make this image make sense and can be used in the daily life.

And further they create knowledge from this map so we can make a decision how we go from

this point to that point.

And this lead to a lot of real world impact for application such as the lift, the Uber

and the maps.

We are asking question can we do this to understand the human body as well because human internal

structure is also very complicated and it's better to be understand to keep the earth

from the disease.

So we use the CT scan as example.

You can see if you are not expert in medical to the high school or graduate student maybe

you have no idea what this image is about.

And if you can annotate the image which one is the organ, which one is the vessel and

the bone and so on.

You pretty much have understanding of what is really going on in my body.

And you can render and make some knowledge from it, make a medical decision, measurement