Distance‐based time series classification approach for task recognition with application in surgical robot autonomy

Abstract

The proposed framework is robust and accurate. Therefore, it can be used to develop adaptive control systems that will be more responsive to surgeons' needs by identifying next movements of the surgeon. Copyright © 2016 John Wiley & Sons, Ltd.

Automated robot‐assisted surgical skill evaluation: Predictive analytics approach

“…New technological innovations such as robotic surgery create great opportunities for automated objective skill assessment and a prompt feedback system, which was not previously available. Robotic surgical devices such as da Vinci (Intuitive Surgical, Sunnyvale, CA) record motion and video data, enabling development of computational models to analyze surgical skills through data‐driven approaches . Techniques such as data mining and machine learning are likely to have a huge impact on ongoing studies of clinical decision support .…”

“…Robotic surgical devices such as da Vinci (Intuitive Surgical, Sunnyvale, CA) 4 record motion and video data, enabling development of computational models to analyze surgical skills through data-driven approaches. 5 Techniques such as data mining and machine learning are likely to have a huge impact on ongoing studies of clinical decision support. 6 The ability of machine learning methods to uncover concealed patterns in a large dataset, such as kinematic and video data, offers the possibility to better understand and model surgical data in order to evaluate surgeons' skill and individualized training.…”

SATR-DL: Improving Surgical Skill Assessment And Task Recognition In Robot-Assisted Surgery With Deep Neural Networks

“…These drawbacks make previous approaches less efficient for an online automatic feedback system. A sequence distancebased method, such as dynamic time warping (DTW), has been proposed to provide feasible online classifying for surgical task and gesture recognition [8]. However, higher computational loads involved in practice, as well as the role of DTW in the skill analysis still remains unknown.…”

Learning‐based automated segmentation of the carotid artery vessel wall in dual‐sequence MRI using subdivision surface fitting

“…For the K‐Nearest‐Neighbor classification, the only parameter that needs to be provided is K. In general, with a small value of K, the classification result is more influenced by noise; whereas a large K value results in a smoothing to the classification. K = 5 has proven to be effective in multiple applications of medical image classification . In the work of Vrooman et al., experiments using K = 1, 10, 45, 100 were carried out to study the influence of K on brain tissue segmentation.…”

“…K = 5 has proven to be effective in multiple applications of medical image classification. 23,24 In the work of Vrooman et al, 25 experiments using K = 1, 10, 45, 100 were carried out to study the influence of K on brain tissue segmentation. Considering that it might be impossible to definitely assign a given candidate to one of the three classes (definitely be lumen boundary (TP), possibly be lumen boundary (PB) and definitely not lumen boundary (TN)) if K is a multiple of 2 or 3, we performed leave-one-out cross-validation in the training set using K = 1, 5, 11, 47, 97.…”