Performance comparison of various feature detector‐descriptors and temporal models for video‐based assessment of laparoscopic skills

Abstract

Temporal modeling of feature histograms extracted from laparoscopic training videos provides information about the skill level and motion pattern of the operator. Copyright © 2015 John Wiley & Sons, Ltd.

Multi‐instance multi‐label learning for surgical image annotation

“…Third, we applied a popular feature representation technique based on speeded up robust features (SURF) features and the bag‐of‐words (BoW) model. Each image was represented by a normalized histogram of visual words as described in Loukas and Georgiou . In the above methods, MLL was based on MLSVM with a Gaussian kernel ( γ = 0.2).…”

An automated skills assessment framework for laparoscopic training tasks

“…To increase the validity of the tests we repeated the K ‐fold experiments 10 times and calculated the average value and standard deviation of the experiments, which are listed in Table . In Table we also present the results of independently applying the process described to the video sequences of each task and the classification results of another computer vision‐based technique that recently showed promising results for skills assessment in laparoscopic surgery …”

“…Assessment of surgical skills based on computational analysis of the recorded video has been introduced as an alternative to sensor‐based kinematic analysis, with significant benefits such as no requirement for additional sensors attached to the surgeon's hand/tools, natural performance of the task and visual analysis of the performance outcomes. For recognition of expertise in laparoscopic training, significant research has been performed on the temporal modelling of salient descriptors extracted from the endoscope video frames . Some works employed internal/external camera combinations or color markers to monitor the trainee's hand movements when performing a task.…”

“…After familiarizing themselves with the equipment, the participants performed two basic laparoscopic tasks on a box trainer: peg transfer and knot tying. The laparoscopic equipment utilized for performing these tasks (Figure ), as well as the tasks themselves (Figure ), have been described in a previous study by the authors . In summary, the first task required placement of four cylindrical pegs into the holes of a pegboard, whereas the second task was more demanding and involved intracorporeal knot tying based on the square‐knot technique.…”

Video Analysis in Basic Skills Training: A way to Expand the Value and use of Blackbox Training?

“…The pegboard was marked with four different markers in each corner as a reference for the video analysis software (Fig. 2) [27].…”