Abstract
Minimally invasive surgery is a surgical intervention used to examine theorgans inside the abdomen and has been widely used due to its effectivenessover open surgery. Due to the hardware improvements such as high definitioncameras, this procedure has significantly improved and new software methodshave demonstrated potential for computer-assisted procedures. However, thereexists challenges and requirements to improve detection and tracking of theposition of the instruments during these surgical procedures. To this end, weevaluate and compare some popular deep learning methods that can be exploredfor the automated segmentation of surgical instruments in laparoscopy, animportant step towards tool tracking. Our experimental results exhibit that theDual decoder attention network (DDANet) produces a superior result compared toother recent deep learning methods. DDANet yields a Dice coefficient of 0.8739and mean intersection-over-union of 0.8183 for the Robust Medical InstrumentSegmentation (ROBUST-MIS) Challenge 2019 dataset, at a real-time speed of101.36 frames-per-second that is critical for such procedures.