A Fundamental Theory for Dexterous Surgical Skills Transfer to Medical Robots
Abstract
Urgent trauma care on the battlefield is a difficult and dangerous thing. Not only is the patient’s life in jeopardy, but the medics, surgeons, and even stretcher carriers face deadly consequences while providing assistance. For over a decade, the military has been looking at technological solutions for a “smart stretcher” and other such technologies with the ability to drag a wounded Soldier out of the line of fire, protect him/her from further injury, and even start the triage and life support operations necessary to prevent further deterioration and begin the recovery process. This proposal does not endeavor to develop such a physical device, but to develop the theories and machine learning platforms necessary to support remote triage and remote semi-supervised surgery portable to “scalably austere environments” (SAE). The reality of such a miracle stretcher is that, despite progress in self-driving cars, autonomous manufacturing robots and ubiquitous cell phones with their sensors and networkability, such a level of autonomy and self-directedness is still mostly in the realm of science fiction. But this proposal employs the known constraints of the operating room and medical lifesaving procedures to develop underlying theory and algorithms necessary to make remote, semi-supervised emergency trauma care a reality. It would be extremely valuable, for example, if the stretcher of the future could simply take a few vital signs during transport in order to prepare the Mobile Army Surgical Hospital (MASH) hospital for the wounds it was about to see. Furthermore, with the prevalence of minimally invasive surgery (MIS) and MIS robots, it has become theoretically possible to have remote surgeons, far from battle, already removing a bullet or clearing a blocked airway during transport of the wounded Warrior to the hospital to buy precious minutes of time. But what is required of the robotics and artificial intelligence (AI) capabilities in order to achieve these life-saving advances that bring care earlier to wounded Soldiers? First and foremost, network lags must be handled in a graceful manner, because if a surgeon is performing “surgery-by-wire” from some safe location away from the battlefield, it is unacceptable for the robotic scalpel on the stretcher or in the automated ambulance to freeze in place or, worse, jerk erratically. We propose methods to interpret “loss of signal” and to take over control in a way that continues the procedure being performed on the patient. But this involves taking over temporarily for the surgeon or medic, so we further propose methods to learn, from a “one-shot demonstration,” what the medical provider is doing based on thousands of previous examples. Fortunately, we only need to learn enough to finish what was started, so we already have a strong idea of what is going on. Still, we must learn this primarily from camera views placed around the patient on the stretcher, which is much less information than the medic, who can touch the patient and feel the scalpel. We can also use the camera views to account for the location of the neck when preparing for a breathing tube or the location of the chest in removing a bullet. This proposed work will not only help speed care to wounded Soldiers in battle, but it will keep medical personnel outside the theater of danger and improve the chances of success for the patient as well as the chances for returning to normal life. These improvements in networked surgery could eventually apply to all types of surgeries in all types of environments because of the context of resource constraints. Imagine a complex surgery needed in a small rural town with little access to specialized care. A doctor from another state, or even another country, that is highly proficient in the needed specialty could provide it from his/her office, without fear of network lags or dropouts. Eventually, new doctors could be trained, remotely, by such techniques, e
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Oct 29, 2018
- Source ID
- W81XWH1810769
Entities
People
- Juan Pablo Wachs
Organizations
- United States Army
- University of Virginia