The Journal of Japanese Society of Stomatognathic Function
Online ISSN : 1883-986X
Print ISSN : 1340-9085
ISSN-L : 1340-9085
PAPERS
Indirect evidence for the reduction of controlled degrees of freedom in chewing
Katsuyuki TakedaYoshinori HattoriTakahisa MurakamiMasaaki IwamatsuMakoto Watanabe
Author information
JOURNAL FREE ACCESS

2010 Volume 16 Issue 2 Pages 102-111

Details
Abstract
Due to difficulties in simultaneous electromyographic (EMG) recording of all the jaw muscles which are involved in chewing, direct evidences of the reduction of controlled degree of freedom (DOF) of chewing, including the information on jaw muscle synergy for this motor task, are scarce. This study aimed to obtain indirect evidence for the reduction of controlled DOF of chewing, by testing the predictability of chewing trajectories from the EMG activities of selected jaw muscles. Because this prediction is possible only when other jaw muscles consistently change their activities simultaneously with the selected muscles, a successful prediction will provide an indirect evidence of synergy, although the inverse may not be true. Methods: The lower incisal path and the EMGs of bilateral masseter, anterior temporalis, and submandibular muscles were recorded during unilateral gum chewing in 4 healthy volunteers. For each of the 200 chewing cycles, the durations of opening, closing, and occluding phases were divided into 10 equal time intervals so as to produce a total of 31 time points. Initially, the incisal path and EMGs in each cycle were described using the 3D coordinates and the rectified EMG amplitudes at the indicated time points, respectively, and then reduced into the data sets comprising the principal component (PC) scores, which account for ≥ 90% of the variations, by a PC analysis. Finally, an artificial neural network (ANN) model, which associates EMGs to incisal path data sets, was trained using 50% of the data sets (learning data) by using a back-propagation learning algorithm, and inspected using the remaining 50% (inspecting data). Results: The ANN model performed best after some training. The incisal paths for the learning and inspecting data sets were predicted successfully. The mean error among the 31 points was 0.7 and 0.8 mm for the learning and inspecting data sets, respectively. The largest errors, 1.5 and 1.8 mm, respectively, were observed during the later opening phase. Conclusion: These findings may provide an indirect evidence of the reduction of controlled DOF in chewing.
Content from these authors
© 2010 Japanese Society of Stomatognathic Function
Previous article Next article
feedback
Top