Inverse Modeling of the OPAL Complex Behaviours

Investigating the highly complex neuromotor control of mastication, swallowing, and related activities forms a key part of our research with of our OPAL model. Neuro-control of these behaviours involve highly coordinated actions between many anatomical components, the disruption of which can cause dysphagia and other disorders. This research will be unique to this project as it requires a high fidelity, integrated biomechanical model of the OPAL complex derived from the imaging and modeling activities.

We are actively looking for graduate students and undergraduates who would like to participate in the activities below. Please contact us if you are interested in getting involved.

Our activity includes:

Estimating Muscle Activation

We study inverse modeling techniques to automatically estimate the muscle activations associated with observed motions of the jaw, tongue, hyoid, palate, pharynx and larynx. A non-linear optimization approach similar to [Sifakis et al. 2005] is used to minimize the error between the actual motions and the predictions of the forward biomechanical model. Combined motion and muscle activity data obtained from facilities available with our associates and partners and from the literature where possible (i.e., [Perlman et al., 1999]) are used to estimate and validate our approach. We also use a Bayesian approach, which allows greater use of prior knowledge and constraints to regularize the estimate.

Motor Control Synergies

It has long been observed that apparently complex voluntary muscle activations can be constructed out of a small set of “muscle synergies” or “motor primitives” [Bernstein 1967, Santello et al. 1998]. Several methods for synergy extraction have been proposed including Independent Component Analysis (ICA) and nonnegative matrix factorization (NMF) [Tresch et al. 2006]. We use these techniques for extracting synergies involved in different categories of movements produced by the OPAL complex. The number of synergies is a hyper-parameter that is optimized in this process.

Modeling of Mastication and Swallowing

We use the above processes to develop a fully-integrated model of mastication and swallowing. We combine our existing muscle-activated models of the jaw-hyoid complex (Hannam, UBC), the ICP tongue (Payan, ICP), and, for comparison, the Dang tongue (Dang, JAIST), with new models of the oropharynx (Hannam, UBC; Miller, UCSF) and larynx (Ludlow, NIH). Experience with multiple muscle inputs in our jaw-hyoid and tongue models has highlighted the need for a more sophisticated way to manage muscle patterning in integrated systems. The completed model requires a food bolus model as well. Normal mastication and swallowing sequences are validated with experimental data including laryngeal movements following electrical stimulation of isolated muscles, peri-laryngeal electromyographic activity, plus jaw, tongue, hyoid and bolus movements during mastication and swallowing.