Automatic endodoscopic image orientation stabilisation with ultra-low-latency

Minimally invasive surgery often makes use of endoscopes with an angled lens. When using such an endoscope, the on-screen image tends to rotate wildly, which may confuse the surgeon. A low-impact way of solving this problem is by using computer vision techniques to track the rotation of the incoming footage and then counterrotating the image. Such an approach has already been proposed in the literature, but it has never been examined whether it is possible to integrate such an algorithm into a state-of-the-art Digital Operating Room environment, where minimal latency is required. In this paper, we compare three different ways of implementing a counterrotation algorithm on the hardware that is available in the NUCLeUS Digital Operating Room of the company eSaturnus.