Michael Kranzfelder1,2, Armin Schneider3, Adam Fiolka3, Sebastian Koller3, Silvano Reiser3, Thomas Vogel3, Dirk Wilhelm4,3, Hubertus Feussner4,3. 1. Department of Surgery, Technische Universität München, 81675, Munich, Germany. michael.kranzfelder@tum.de. 2. Research Group MITI (Minimally invasive Interdisciplinary Therapeutical Intervention), Klinikum rechts der Isar, Technische Universität München, 81675, Munich, Germany. michael.kranzfelder@tum.de. 3. Research Group MITI (Minimally invasive Interdisciplinary Therapeutical Intervention), Klinikum rechts der Isar, Technische Universität München, 81675, Munich, Germany. 4. Department of Surgery, Technische Universität München, 81675, Munich, Germany.
Abstract
PURPOSE: Laparoscopic cholecystectomy is a very common minimally invasive surgical procedure that may be improved by autonomous or cooperative assistance support systems. Model-based surgery with a precise definition of distinct procedural tasks (PT) of the operation was implemented and tested to depict and analyze the process of this procedure. METHODS: Reliability of real-time workflow recognition in laparoscopic cholecystectomy ([Formula: see text] cases) was evaluated by continuous sensor-based data acquisition. Ten PTs were defined including begin/end preparation calots' triangle, clipping/cutting cystic artery and duct, begin/end gallbladder dissection, begin/end hemostasis, gallbladder removal, and end of operation. Data acquisition was achieved with continuous instrument detection, room/table light status, intra-abdominal pressure, table tilt, irrigation/aspiration volume and coagulation/cutting current application. Two independent observers recorded start and endpoint of each step by analysis of the sensor data. The data were cross-checked with laparoscopic video recordings serving as gold standard for PT identification. RESULTS: Bland-Altman analysis revealed for 95% of cases a difference of annotation results within the limits of agreement ranging from [Formula: see text]309 s (PT 7) to +368 s (PT 5). Laparoscopic video and sensor data matched to a greater or lesser extent within the different procedural tasks. In the majority of cases, the observer results exceeded those obtained from the laparoscopic video. Empirical knowledge was required to detect phase transit. CONCLUSIONS: A set of sensors used to monitor laparoscopic cholecystectomy procedures was sufficient to enable expert observers to reliably identify each PT. In the future, computer systems may automate the task identification process provided a more robust data inflow is available.
PURPOSE: Laparoscopic cholecystectomy is a very common minimally invasive surgical procedure that may be improved by autonomous or cooperative assistance support systems. Model-based surgery with a precise definition of distinct procedural tasks (PT) of the operation was implemented and tested to depict and analyze the process of this procedure. METHODS: Reliability of real-time workflow recognition in laparoscopic cholecystectomy ([Formula: see text] cases) was evaluated by continuous sensor-based data acquisition. Ten PTs were defined including begin/end preparation calots' triangle, clipping/cutting cystic artery and duct, begin/end gallbladder dissection, begin/end hemostasis, gallbladder removal, and end of operation. Data acquisition was achieved with continuous instrument detection, room/table light status, intra-abdominal pressure, table tilt, irrigation/aspiration volume and coagulation/cutting current application. Two independent observers recorded start and endpoint of each step by analysis of the sensor data. The data were cross-checked with laparoscopic video recordings serving as gold standard for PT identification. RESULTS: Bland-Altman analysis revealed for 95% of cases a difference of annotation results within the limits of agreement ranging from [Formula: see text]309 s (PT 7) to +368 s (PT 5). Laparoscopic video and sensor data matched to a greater or lesser extent within the different procedural tasks. In the majority of cases, the observer results exceeded those obtained from the laparoscopic video. Empirical knowledge was required to detect phase transit. CONCLUSIONS: A set of sensors used to monitor laparoscopic cholecystectomy procedures was sufficient to enable expert observers to reliably identify each PT. In the future, computer systems may automate the task identification process provided a more robust data inflow is available.
Authors: Thomas Neumuth; Pierre Jannin; Juliane Schlomberg; Jürgen Meixensberger; Peter Wiedemann; Oliver Burgert Journal: Int J Comput Assist Radiol Surg Date: 2010-06-06 Impact factor: 2.924
Authors: Martin Wagner; Andreas Bihlmaier; F Mathis-Ullrich; B P Müller-Stich; Hannes Götz Kenngott; Patrick Mietkowski; Paul Maria Scheikl; Sebastian Bodenstedt; Anja Schiepe-Tiska; Josephin Vetter; Felix Nickel; S Speidel; H Wörn Journal: Surg Endosc Date: 2021-04-27 Impact factor: 4.584
Authors: M Berlet; T Vogel; D Ostler; T Czempiel; M Kähler; S Brunner; H Feussner; D Wilhelm; M Kranzfelder Journal: Int J Comput Assist Radiol Surg Date: 2022-05-28 Impact factor: 3.421