Today is the Day 1 of Satellite Events at MICCAI 2024 and we welcome the first MICCAI Workshop on Embodied AI and Robotics for Healthcare (EARTH). It will be held in the Borj 2 meeting room from 6th October - PM.
We interviewed Zhen Chen, an Assistant Professor at the Centre for Artificial Intelligence and Robotics at the Chinese Academy of Sciences in Hong Kong who is a co-organiser of EARTH Workshop to learn more about his research and the purpose of the workshop.
Embodied artificial intelligence is in essence an AI that has a physical presence which interacts with the real world. For surgery, I think research about embodied AI can be formulated like a three-level pyramid.
At the foundation level, basic AI techniques are crucial. AI can analyse surgical videos, identify different instruments, and recognize procedure stages. We can also use this to help train new surgeons.
The middle level of the pyramid is surgical guidance. When an AI is more powerful and accurate, it can be used during surgery to perform real-time analysis, like alert surgeons to potential risks or suggest optimal instrument placement.
On the top of the pyramid, embodied AI is like a physical surgical assistant. It can perform tangible actions like manipulating an endoscope based on voice commands. This is particularly useful because the surgeon has two hands but several tools. We want an embodied AI that can listen to the surgeon, understand the intention, and operate the endoscope, freeing up the surgeon's hands. So, embodied AI is like a pyramid; we are building it from the bottom to the top.
It's a really exciting time in this field right now. A few things stand out to me: First, we are seeing AI and robotics experts working together a lot more. This is important because the big challenge in embodied AI for surgery is getting smart software to work smoothly with advanced hardware. It's not just about having a clever AI or a precise robot; it's about making them work together seamlessly in the complex environment of an operating room. This collaboration is driving innovation and helping us solve problems that neither field could tackle alone, like our centre.
Second, we are noticing this field is getting attention from different research communities. Over the past two years, we have organised a workshop at IJCAI [International Joint Conference on Artificial Intelligence], and one at ICRA [International Conference on Robotics and Automation] this year. Now, we're bringing this discussion to MICCAI for the first time. This cross-pollination of ideas is great.
Lastly, I think we are at a tipping point where research is turning into practical applications. We are seeing more clinical trials and actual use of AI-assisted surgical robots which is driving further research.
The unique advantage of MICCAI is its audience. For example, at IJCAI, the audience mainly focuses on AI, and for ICRA, the audience focuses on robotics. But actually, we are working on embodied AI to address a healthcare problem. This is the advantage of MICCAI: our audience are experts in healthcare. They are more knowledgeable about applications, for example, how different AI and robotic techniques can be applied in different scenarios in the surgery.
Actually, my PhD focused on computer-aided diagnosis. For example, when a patient comes to a hospital, they will get an examination, like x-rays, histopathology, and even an endoscopic examination. And we'll use algorithms to analyse the patient and that will help the clinicians. But that's not the end of the healthcare pipeline for the patient. Maybe this patient needs some medical intervention, for example, a surgery. So, after my PhD, I wanted to research the rest of the pipeline. I accepted the assistant professor position at the CAIR in Hong Kong where I have colleagues from robotics because I found that, for surgery, we can really combine our advantages together.