In mid-October, 150 surgeons, technologists, venture capitalists, and academics came together to discuss critical issues for surgery, neurosurgery, AI, and medicine. The four-day Digital Neurosurgery conference in Palo Alto brought together the worlds of technology and surgery to break down silos and forge new connections in the world of artificial intelligence for surgeons.
“We wanted to move the field forward in the digital area, across silos, fields, and types of people," says Daniel Donoho, founder of the Surgical Data Science Collective (SDSC), who co-organized the conference. “We exceeded any reasonable expectations we could have had.”
There's a huge opportunity around data in the operating room – but extracting data from the visual scene has been a gap in surgeons’ knowledge, says Dr. Samuel Browd, who co-organized the conference with Donoho. “The ability to capture intraoperative imaging, do scene segmentation, archive this data and learn from and share this information is going to profoundly change how we operate, how we teach, and how we do research,” he says. “There's a fundamental moment in time occurring where there's a convergence of technologies that will allow this to happen at a broad scale, and really democratize surgery across the world.”
Sessions at the conference included talks on what surgeons can learn from pro sports analytics, the future of technology in surgery, and how machine learning may change the way neuroimaging works. Margaux Masson-Forsythe, Director of Machine Learning with SDSC, presented an early workshop on how computer vision is used in many fields including climate tech and manufacturing.
Masson-Forsythe also spoke on computer vision to help analyze videos captured of surgeries– specifically, removing a tumor from a pituitary gland endoscopically. She showed the audience the step-by-step method of creating a dataset to train a model to detect, identify, and track multiple surgical instruments in the surgery, and how to optimize the model.
It’s all part of SDSC’s goal to take the hours of operative videos that surgeons have on file and transform them into datasets built for machine learning that can be used to analyze the videos and export analytics. SDSC can help with a platform that has 3,700 hours of surgeries already uploaded – with the idea that artificial intelligence can help improve surgeons’ accuracy and patients’ outcomes.
Surgical data science and machine learning systems are on the cusp of transforming the field of surgery. By augmenting the skills and expertise of surgeons, these systems offer the potential to make procedures safer, more efficient, and ultimately improve patient outcomes. As technology continues to advance and regulatory frameworks adapt to this changing landscape, the integration of machine learning into surgical practice will become increasingly common.
Masson-Forsythe says there are still open questions about the permissions and data uploading of the surgeries – but the meeting created the conditions to move forward in a productive way. Computer vision experts usually go to conferences with other vision experts, and surgeons might attend conferences on specific surgeries. But this meeting was a different situation, she says, and it occasionally required different groups to translate what they were saying in order to communicate with peers. “Bringing everyone together like this in that conference really started to create some sparks of innovation and collaboration between a lot of different people that would have never met otherwise.”
Donoho agrees, adding the number of people who have reached out after the conference is unprecedented. “I’ve heard people talking about starting companies or research projects together – it has been a joy to see the outcomes.”
The next steps for SDSC, Donoho says, are to use the momentum of the conference to continue to acquire new users and partners within the surgical community. “There is a tremendous outpouring of people who would say: if I only had this system, then I could do incredible things with it,” he says. “Now, our fundamental task is to remind them that we're out there, we exist, and we can do incredible things collaboratively.
.png)
