Pupil Variation Applied to the Eye Tracking Control of an Endoscopic Manipulator

In laparoscopic surgery, numerous devices have been developed to allow surgeons to manipulate the laparoscope by themselves. Some previously adopted approaches include hands-free strategies, such as eye tracking. In this letter, we propose a new approach for the control of an endoscopic manipulator using pupil variation, which has not been previously attempted. We developed an intention recognition system for an endoscopic manipulator based on a support vector machine (SVM) and a probabilistic neural network (PNN). The SVM classifier, trained on pupil variation and eye rotation velocity data, recognizes when the operator wants to alter the direction of the endoscope. The PNN classifier determines in which direction the operator wants to move. We set up an experimental task to evaluate our proposal, and conclude that pupil variation has a significant effect on judging the timing for activating the endoscopic manipulator to project the operative field onto the center of the visual field on monitor. Moreover, it shows better performance than endoscope manipulation by an assistant.