PublicationDetail

Chair for Computer Aided Medical Procedures & Augmented Reality
Lehrstuhl für Informatikanwendungen in der Medizin & Augmented Reality

THIS WEBPAGE IS DEPRECATED - please visit our new website

T. Sielhorst, T. Blum, N. Navab
Quantitative comparison of movements for medical training
4. Jahrestagung der Deutschen Gesellschaft für Computer-und Roboter-Assistierte Chirurgie (CURAC 2005), Berlin, Germany, September 2005 (bib)

Purpose:

In CURAC 2004 we proposed an Augmented Reality System for medical training. It consists of augmented reality system and a medical simulator for training. A valuable addition to such a training system is an appropriate way of synchronizing the movements of varying speed in order to simultaneously visualize the motion of experts and trainees, and to study trainees performances quantitatively.

Methods:

In order to compare two performances of the same action we want to
• be able to visually compare two trajectories either done by professionals or by a student and a professional. For this purposes we want to replay two previously performed and recorded motions synchronously, using AR to have an omnidirectional viewing on both which helps to identify and study subtle differences.
• get a similarity measure between two previously recorded trajectories to quantitatively measure and automatically judge the performance of a student who tried to reproduce the movement of a professional Straight forward approaches like a global speed adjustment, feature matching or closest neighbor matching do not provide appropriate results. The most promising algorithms for comparing trajectories were investigated for the task. These are namely the LCSS (longest common subsequence) and the DTW (dynamic time warp).

Results:

We discovered that finding a time-invariant similarity measure between two 3D trajectories and a synchronization in time of both can not be solved using simple approaches. However, there are problems which use mathematically similar models like speech recognition or signature verification, and can be solved using Dynamic Time Warping or Longest Common Subsequence. We implemented both methods and evaluated them for our application. LCSS is able to provide a similarity measure and a mapping that synchronizes one trajectory to the other. Problems appear only if both movements have been recorded using different update rates. DTW is also capable of providing a similarity measure and a synchronization. Problems with different update rates can be overcome by a minor change. Experimental results show that the system provides appropriate synchronization and allows us to quantitatively compare similar tasks executed by different users at different speeds.

Conclusion:

In our work, we present systems and methods which allow us to synchronize and compare sequences of captured 3D movements. The method is applied to an AR system designed for training of physicians and midwives. One of the algorithms gives us a similarity measure that could be used to automatically rate the performance of a trainee when trying to replicate a movement. In fact this similarity measure can currently not replace the supervisory comments of a professional. The whole action is very complex. It has parts of different importance. An experienced supervisor knows the crucial parts and can include his knowledge into his judgment of trainees performance. Nevertheless the warping path that can be obtained when computing the similarity measure is very valuable, as it gives us the possibility to visualize both movements simultaneously and have a look at the differences between two actions. Since we use augmented reality to replay the synchronized movements, we are able to examine them from different viewpoints and analyze them in a much more tangible way than video recordings would offer. Also our implementation allows to change the speed at which the two trajectories are shown, stop them or rewind the replay through which a very detailed analysis of the trajectories is possible.
This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each authors copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.



Edit | Attach | Refresh | Diffs | More | Revision r1.13 - 30 Jan 2019 - 15:16 - LeslieCasas

Lehrstuhl für Computer Aided Medical Procedures & Augmented Reality    rss.gif