Welcome back to Beyond the Patterns.

So today I have the great pleasure to introduce Xiaoxiong Lai.

So he has been working with photo and counting CT and in particular with the ASICS technology

on the detector side.

And he will present today on how to combine the knowledge about the detector, the ASICS

and on the other hand the physics and deep learning into one system and demonstrate how

this can give you much, much better images for photo and counting CT.

So it's a great pleasure that we can now start with another episode of Beyond the Patterns.

Okay, so welcome everybody.

Today I have the great pleasure to introduce Xiaoxiong Lai.

He is from Shanghai Tech.

At his bachelor's in China from the University of Science and Technology, then moved to the

University of Illinois, Champaign and did his PhD focusing on energy-resolved photo

and counting detectors and SPECT MRI.

He is also a Gordon Research Fellow at Harvard Medical School and MGH and followed up there

for static CT for space missions and cardiac image reconstruction.

Then he went on to be a senior scientist at Canon Medical Research working on photo and

counting detectors for SPECTRO CT.

And in 2021, he joined Shanghai Tech University as tenure track professor leading the Photonic

Sensing and Imaging Group, the PSI Lab, in the School of Biomedical Engineering.

He specialized in advanced semiconductor detectors, electronics and imaging systems for biomedical

applications and recently focused, of course, on photo and counting CT.

Today's talk will be entitled a physics, ASICs architecture driven deep learning model for

photo and counting CT.

And it's a huge pleasure that you're here.

Xiaoxiong, the stage is yours.

Thank you.

Thank you for Andrew's introduction.

It's a great honor.

I'm also very excited to share our recent work regarding using the physics ASIC architecture

driven deep learning model.

Actually, it's a detector model for photo and counting CT.

So this one is a joint effort between us and also the Southeast University of China.

Here is the disclosure.

So this project is funded and supported by United Imaging Healthcare in China and also

Shanghai Technology Commission of Shanghai Newspaper Project.

Here is the founding ID.

And so the overall line of my talk will give a brief introduction of the photo and counting

CT and include CT basic effects and also ASIC comparison, application and challenges,

and related work.

Then introduce our physics ASIC architecture guided deep learning model for photo and counting

CT and we show some results or preliminary results.

And for CT, I think everyone regarding is a basic physics review.

So actually you have a tube, the electron being hitting your rotating anode.

So you get because the electron is speed up, have high energy, then hitting is stopping

at the same time will eliminate a lot of x-ray.

Those x-rays are multi-special, they have a different energy.

Then those x-rays are hitting with our object, hitting lowest patients.

Then because the interaction with our photoelectric effect and complex scattering actually would

reduce if you have a high attenuation tissue like bone, then you detect less photons.