Theator Files for IPO: AI-Powered Surgical Video Platform Aims to Enhance Surgical Outcomes

AI Is Teaching Cars to Make Better Decisions—Can It Help Surgeons, Too?

Solving this problem is Theator’s mission. Theator, a startup based in Palo Alto, California, with an R&D center in Tel Aviv, is striving to drive the nascent revolution in autonomous surgery.

Dotan Asselmann, Co-founder and Chief Technology Officer of Theator, stated that his company has been monitoring the progress of autonomous vehicles and using them as a blueprint for surgical procedures, with a focus on leveraging AI-driven analytics to enhance decision-making capabilities.

Just as autonomous vehicle manufacturers aim to halt vehicles before accidents occur, Theator seeks to stop surgeries before any errors arise. This technology is achieved by analyzing surgical videos from around the world.

Asselmann stated, “Due to its scalability, artificial intelligence can accumulate more experience than any surgeon. Our model has analyzed thousands of surgical cases, whereas an individual physician simply does not have the time to perform such a large volume of surgeries.”

Asselmann and Theator's team found thatThe Lack of Standardization in Surgical Examination Procedures.Asselmann stated that while most surgeons learn their craft from a select few mentors, the majority of their knowledge is derived from their own clinical experience.

“Horizontal data sharing among surgeons has long been limited—primarily confined to conferences,” he said. “Due to COVID-19, surgeons’ ability to scale up knowledge dissemination has been stifled.”

However, although the practice of visually assisted surgery has begun and most operating rooms are equipped with cameras to record surgical procedures, surgeries are not routinely captured, stored, or analyzed. It is this gap that inspired the founding of Theator and continues to drive its mission to empower surgical procedures through artificial intelligence and computer vision.

The company’s technology is implemented via edge devices mounted on laparoscopic carts in operating rooms. The NVIDIA Jetson AGX Xavier platform processes the video, after which Theator’s software anonymously uploads it to its training environment hosted on Amazon and Azure clouds.

The company runs various AI models, training them on clusters of NVIDIA V100 Tensor Core GPUs, which also handle inference.

Once a surgical video has been processed, surgeons can immediately focus on reviewing highlighted data packets at selected critical moments, thereby making important decisions. Each procedure is added to Theator’s training dataset, expanding the understanding of its models.

By applying AI-driven analytics to video, Theator’s platform breaks down outcome data into steps, events, decisions, and milestones. This enables surgeons to conduct postoperative reviews and compare segments of a procedure with those from previous similar surgeries.

The platform can also leverage historical surgical data to provide preoperative assistance, and it can match video footage to identify the causes of postoperative complications. Future applications include the ability to predict, and potentially reduce, costly and time-consuming interventions resulting from such complications.

For example, a patient with postoperative fever may have an uncontrolled bleeding site. In the future, watching video summaries on television could help surgeons determine whether problems exist before proceeding with scans or corrective surgery.

Asselmann believes that Theator can eliminate cloud components within one to two years, achieving the holy grail of real-time surgical support—performing the entire analysis process in minimally invasive surgery relying solely on its artificial intelligence algorithms.

Although the company’s current focus is on assisting surgeons, he anticipates that semi-autonomous surgery could become a reality within the next five years. While surgeon involvement may still be required, Asselmann believes that Level 3 or Level 4 surgical automation will first be adopted in developing countries, where 5 billion people lack access to adequate surgical resources.

With the help of the NVIDIA Inception program, Theator has achieved its current success. This is an accelerator program for startups in the fields of artificial intelligence and data science. Asselmann praised the program for helping to “improve our model training efficiency, reduce computational costs, and guide the proper hardware selection for edge devices.”

Through the NVIDIA Inception program, Theator also gained private access to demonstrations of the NVIDIA Clara Guardian AI healthcare framework and the NVIDIA DeepStream SDK, which the startup used to build its high-efficiency, real-time video pipeline.

With NVIDIA’s support, Theator can continue to help surgeons worldwide make decisions in the operating room.

Asselmann stated, “Surgeons are overwhelmed by an endless stream of parameters from multiple directions during surgery.” “Our goal is to reduce cognitive load and help them make optimal decisions for the right patient and situation at the right time.”

“You can still call the shots,” he said, “but with artificial intelligence, you will become a better surgeon.”