gms | German Medical Science

Background: Bariatric surgery is the recommended treatment option for patients suffering from morbid obesity. Out of the existing bariatric procedures, Laparoscopic Roux-en-Y gastric bypass (LRYGB) is the most commonly performed. However, only about 53% of bariatric surgeons measure limb length. This is especially important as loss of depth perception is one of the main drawbacks of laparoscopic surgery. With new 3D-imaging technology depth perception has improved, yet it remains challenging to objectively measure distances during laparoscopic procedures as in LRYGB. To address this challenge, we developed and evaluated a computer-assisted bowel length measurement system (BMS), which adds measurement functionality to conventional 3D-laparoscopes and thereby allows for quantitative laparoscopy.

Materials and methods: BMS was used with a 3D-laparoscope (3D TipCam 1, Karl Storz GmbH, Tuttlingen, Germany). Laparoscopic measurement of bowel length was conducted in four steps. After the measurement command was given via USB-footswitch, computer vision was employed to segment bowel and instrument tips in the stereoscopic images (1). From the segmented images a 3D-point-cloud was created and transformed into a triangular mesh (2). The shortest distance of the instrument tips was then measured alongside the triangular mesh (3) and the result displayed to the surgeon as augmented reality (4).

For thorough preclinical evaluation, phantom (n=50, two users), ex-vivo-porcine (n=50, two users) and in-vivo porcine trials (n=10, five users) were performed. In all experiments a total distance of 70cm was measured with BMS. Afterwards ground truth was obtained by manual measurement of the bowel, requiring laparotomy during the in-vivo porcine trials. Accuracy was calculated as ratio of ground truth to length measured with BMS.

Results: Mean value measured in phantom trial was 66.1 cm (standard deviation (SD) 2.7 cm) and accuracy of 94.43%. Ex-vivo-porcine trial resulted in a mean value of 65.8 cm(SD 2.55 cm) and accuracy of 94.03%. In-vivo-porcine trials a mean value of 67.5 cm with a standard deviation of 6.8 cm and an accuracy of 96.43 % was achieved.

Usage within a clinical setting in a human patient was feasible. A total of 17 measurement commands were given. The first 5 measurements failed because of a third instrument in the laparoscopic image. After removal of the third instrument 11 out of 12 measurements were successful. With a measurement goal of 60 cm, the system measured 43,4cm (72,33%). Here no ground truth was obtained, but the surgeon estimated the bowel length with marks on the instruments. The system provided adequate measurement feedback and was well integrated into surgical workflow.

Conclusion: BMS enables safe and precise laparoscopic bowel length measurement. By a stepwise evaluation and iterative development it was possible to translate the system from bench to bedside. It ads functional value to conventional 3D-laparoscopes and is well integrated into surgical workflow. As a first application of quantitative laparoscopy it could allow for more standardized bariatric surgery.

Acknowledgements: This research has been funded by the German Research Foundation DFG within the Transregional Collaborative Research Center 125 “Cognition-Guided Surgery”, project A01 and by the Medical School of Heidelberg University with a Physician-Scientist-Fellowship for Martin Wagner.