Robot Learns Wound Stitching

Various doctor-assisted surgeries including hernia repairs and coronary bypasses have been supported by robots for quite some time now. However, these surgeries still involve the surgeon as an important part of the process. Now, latest research outlines the potential of using more autonomous robots in procedures that require intricate, detailed tasks such as suturing. The development of this research could have potential applications in other sectors of robotics as well.

Robotic researcher Ken Goldberg from UC Berkeley, who directs the lab where the research was conducted, agrees that this procedure is quite a feat from a robotic point of view. Bright, reflective objects such as needles can confuse the robotic image sensors. In addition, it is quite difficult for computers to gauge and mimic the behavior of pliant objects like skin and thread when they are touched or pricked. Unlike passing a needle between two human hands, it poses a greater challenge for robotic arms due to the dexterity involved.

The robot makes use of two cameras to analyze its surroundings. Following which, it identifies the needle, thanks to the training provided on a neural network. The robot also has a motion controller that plans all the six movements required to stitch.

Though the employment of these robots in operating rooms is quite far from reality, their use in automating the suturing process can offer significant medical benefits. Danyal Fer, a physician and researcher involved in the project, says that suturing is usually the final task of a surgery procedure. This makes the doctors more susceptible to fatigue, which can result in improper suturing and lead to delayed healing and a range of other complications. Considering the repetitiveness of suturing, both Goldberg and Fer view it as an ideal task for automation.

Goldberg says, “Is there a possibility to prove that this improves patient outcomes? Besides being convenient for the doctor, can it lead to improved suturing, faster healing, and reduced scarring?”

However, a number of challenges arise with the use of robots. For instance, the success rate of the robot is still under question. The best the robot has achieved so far is six complete stitches before it required human intervention. On average, it was able to complete about three stitches in the trials. The experimental wounds were also limited to two dimensions, unlike the wounds on certain body parts like the elbow or knuckle which are much more complex. Moreover, the robot was also only tested on phantoms, a type of artificial skin used in medical training, and not on organ tissue or animal skin.