Keynote: Generative AI in Medical Imaging Just for Fun or the Game Changer?
Event: SCBX Unlocking AI EP4, Computer Vision: How AI See Things Like We Do
Collaboration: SCBX and Insiderly.ai
Venue: SCBX NextTech, Siam Paragon, 4th Floor
Speaker: Dr. Piyarit Ittichaiwong , CEO & Co-Founder, PreceptorAI
Today, we can use generative AI to help humans do a variety of tasks, from paperwork, accounting, translation, to creative work. Not only that, AI today also has medical benefits, with many doctors committed to using it to treat patients effectively. Upgrading the country's health sector
One of them is Dr. Piyarit Ittichaiwong. Dr. Siriraj Hospital, who is also the CEO and Co-Founder of PreceptorAI technology company, gave a lecture on the topic "Generative AI in Medical Imaging: Just for Fun or the Game Changer?"
Dr. Piyarit recalled that in the past, when doctors would diagnose specific diseases such as blood cell analysis. You have to take doctors who train specifically in this field, but with the number of doctors who are trained, there are not many doctors. If there is a patient who needs to be treated in a timely manner, he will not be able to take care of him in time. Therefore, I tried to find out if there was a way to help patients faster. This has led to a new generation of patient care innovations that can not only help analyze blood cells, but also analyze parasites and many others.
Dr. Piyarit said that there is a very popular generative AI program that occurred in 2014 by Google to create a newly developed image in the GAN or Generative Adversarial Network system. Thieves vs Police
In other words, it is a technology to create a sharper image that is like making counterfeit banknotes as realistic as possible. The thieves not only counterfeit the banknotes as if they were one-sided, but also have a model who acts as a police officer to detect whether the banknotes that come out are real or fake. Therefore, both parties must always improve themselves so that the other party does not detect it.
That's the principle that enables the creation of the most realistic AI images, and how they can be used to create CT scans.
However, There are many models that have been developed to create images with AI, but the most effective one is the GAN system, and the so-called Diffusion Models have emerged that can create a realistic image that cannot be distinguished from a real image or a generated image.
However, X-ray images that are often encountered in medicine often have a lot of noise. Such a principle will help to create a perfect new image with a technique. "Sprinkle sand" by letting the AI learn from real images that gradually Sprinkle sand on the picture so that it is not clearly visible.
Then, they learned by "sucking out the sand", making the X-ray image invisible clearly. Gradually, it became clearer, more perfect, and became the most realistic picture. The type that can see whether the X-ray film shows an abnormal lump or not. And it is very realistic, the kind that many doctors themselves still can't distinguish between the images generated by AI. Which picture is the real picture?
The tangible benefit is that it helps AI generate lifelike images. It helps doctors not have to wait to learn from the real thing, but they can learn at all without waiting for the actual patient.
It is true that many people may think that AI helps create images, and many people may use it for fun and funny, but in reality, there are many people and professions who use it to create greater benefits. Doctors from Siriraj Hospital believe that this is a technology that will be a game changer to help the medical community and lead to the ability to grow in medicine in the future. It will help patients a lot and turn Thailand into a better country to live in.
