Artificial Intelligence in Intraoperative Surgical Guidance
Published by: Harsimran Singh, Undergraduate at UC Berkeley
Imagine undergoing a five-hour-long craniotomy to have a tumor precisely removed. The neurosurgeon comes in to tell you that most of the tumor was successfully removed with the help of Artificial Intelligence (AI). You are struck with awe. The concept of an advanced self-thinking computer helping identify your tumor and tracking every important blood vessel and nerve, while simultaneously helping your physician know where to maneuver their blade, seems unreal. Nonetheless, this is quickly becoming the reality of the advanced technological world we live in. AI plays a unique role in various subfields of the medical industry such as forming diagnoses, conducting imaging, producing treatment plans unique to their respective patients, enhancing bedside manners, robotic assisted surgery, and much more (3). However, AI can also play an important role in intraoperative guidance during surgery. Intraoperative guidance is how surgeons can be assisted by AI as they perform real-time surgery.
Machine Learning: Supervised vs. Unsupervised
Intraoperative surgical guidance is executed from an AI process called Machine Learning (ML). This is a skill that allows AI to learn so that it can classify different anatomical structures and realize hidden patterns during surgery (2). A surgeon can directly feed information into the ML algorithm (called Supervised learning). During a subsequent surgical procedure, the AI will remember without any additional input from the surgeon. For example, when a surgeon is operating in someone’s abdomen, the ML system will learn to identify whether a particular structure is a gallbladder or not because it has previously learned what a gallbladder looks like from the surgeon. Afterwards the AI system will be able to guide the surgeon by telling them to either make an incision using a “Yes-GO” signal (depicted as green) or to not make an incision via a “No-GO” signal (depicted as red) (1). This provides a valuable framework for the surgeon to know where to avoid using his blade/instrument and when to dissect. The AI can also undertake unsupervised learning, in which it learns to identify hidden patterns on its own without human input. Using the gallbladder example, the AI machine can learn to differentiate between bleeding and non-bleeding sites, which could indicate a possible injury to the gallbladder (1). The surgeon will then receive this information and be able to configure where the wound site is located, allowing for improved navigation during the procedure.
Machine Learning: Computer Vision
Computer vision is a unique feature of ML that is defined as helping machines to understand how to process images, videos, and to differentiate between objects and scenes. This results in machines achieving human-level capabilities that will ultimately serve to increase positive surgical outcomes (4). For example, real-time analysis of laparoscopic video yielded 92.8% accuracy in a surgical gastrectomy, which is a procedure to remove part of the stomach. Even in a complex surgical procedure, AI still is able to yield high accuracy in knowing what parts of the stomach to incise (5). Computer vision which improves ML for AI is very effective at identifying hidden patterns in recorded surgeries that may be imperceptible to humans by employing techniques such as data regression and analyzing valuable relationships/patterns from recorded surgeries. For example, it has executed logistic regression for prediction of surgical site infections (SSI) by creating models that incorporate multiple data sources, including diagnoses, treatments, laboratory values, and previous surgical outcomes (6).
Conclusion
Overall, AI is quickly rising to its way in the healthcare industry. AI uses ML to help guide surgeons how make precise incisions and identify hidden patterns such as bleeding or important nerves and other the anatomical structures. Together, surgeons and the ML system of AI will play an essential role in enhancing surgical outcomes and accuracy, which is evident in the already successful accuracy of surgical outcomes and growing data that supports AI use.
References
Hashimoto DA, Rosman G, Rus D, Meireles OR. Artificial Intelligence in surgery: Promises and perils. Annals of Surgery. 2018;268(1):70-76. doi:10.1097/sla.0000000000002693
Deo RC. Machine learning in medicine. Circulation. 2015;132(20):1920-1930. doi:10.1161/circulationaha.115.001593
Alowais SA, Alghamdi SS, Alsuhebany N, et al. Revolutionizing Healthcare: The role of Artificial Intelligence in Clinical Practice. BMC Medical Education. 2023;23(1). doi:10.1186/s12909-023-04698-z
Moglia A, Georgiou K, Georgiou E, Satava RM, Cuschieri A. A systematic review on Artificial Intelligence in robot-assisted surgery. International Journal of Surgery. 2021;95:106151. doi:10.1016/j.ijsu.2021.106151
Volkov M, Hashimoto DA, Rosman G, et al. IEEE International Conference on Robotics and Automation. Singapore: 2017. Machine Learning and Coresets for Automated Real-Time Video Segmentation of Laparoscopic and Robot-Assisted Surgery; pp. 754–759.
Soguero-Ruiz C, Fei WM, Jenssen R, et al. Data-driven Temporal Prediction of Surgical Site Infection. AMIA Annu Symp Proc. 2015;2015:1164–73.
