Objectives

The Twente Lower Extremity Model (TLEM) dataset is currently based on measurements of one cadaver body. In order to predict other subjects’ functionality, anthropometric scaling based on the subject’s height and weight has been used in the past. Although this is commonly used, it is a relative crude method. With the functional measurements of healthy subjects, including maximal voluntary contraction trials, as performed in Work Package 1, and the 3D parameterization of the MRI images, as performed in Work Package 2, the TLEM model implemented in the AnyBody Modeling System can be scaled to subject specific models, resulting in better prediction of subject specific functional outcome.

Progress

Sensitivity analyses of functional measurements and MRI-parameterization errors

Accurate predictions of subject-specific musculo-skeletal models are essential in TLEMsafe project to obtain reliable estimation of functional outcome after the surgery. Clearly, many parameters affect the predictions of a musculo-skeletal model, but their effect may vary from muscle to muscle. Therefore, it remains unclear which information needs to be collected with greater detail, and what is the required accuracy of the measurement techniques.

For these reasons, an extended sensitivity analysis was performed to quantify the effect of potential errors on medical imaging and functional measurements on the subject-specific model outcomes. The results of this study provided a priority list containing quantitative information about which parameters and which muscles need to be selected most carefully, through accurate measurements and detailed optimization, to create a reliable subject-specific musculoskeletal model, and provided suggestion for measurement protocols and new methodologies to be developed in WP1 and WP2 for obtaining accurate estimates in vivo (Figure 1-2).

Figure 1 - Sensitivity of subject-specific musculo-skeletal models to potential errors in musculo-tendon architecture. TLEM model showing all the muscle-tendon actuators (left) and the most sensitive actuators to perturbation of muscle-tendon parameters (right).

Figure 2 - Sensitivity of subject-specific musculo-skeletal models to potential errors in musculo-skeletal geometry. TLEM model showing all the origin, insertion and via points (left) and the most sensitive points to geometrical perturbations (right).

Subject-specific musculo-skeletal model

Starting up from the measurements of the healthy subject performed in WP1, a first basic version of subject-specific model was made using, as initial method, simple linear scaling laws based on subject’s height and weight. The simulations were based on a series of daily living tasks performed by the healthy subject during the measurements in WP1: gait at slow, normal and fast speed, beginning and termination of gait, stepping over an obstacle, stepping on/off a block, getting up and sitting down a chair (Figure 3). Nevertheless, anthropometric scaling is quite crude, and the results were not yet accurate.

Figure 3 - Subject-specific model performing various daily living tasks: stepping over an obstacle (left), stepping on a block (center), getting up from a chair (right).

Improved versions of the subject-specific model were obtained scaling musculo-tendon architecture (maximal muscle force, optimal muscle fiber length, tendon slack length) to the completely segmented MRI scan (muscle volumes) developed in WP2, and to the maximal voluntary contraction trials (maximal isometric and isokinetic muscle forces) measured in WP1, to reflect subject’s specific force generating characteristics (Figure 4).

Figure 4 - Functional scaling of subject-specific musculo-tendon architecture (center). Muscle volumes are based on complete manually segmentated MRI scan (right), while tendon slack length and optimal muscle fiber length are optimized in order to reflect subject-specific force generating characteristics as measured during the maximal voluntary contraction trials (left).

Next versions of the subject-specific models will be even more accurate, scaling musculo-skeletal geometry to the specific functional measurements (joint axes and centre of rotation) performed in WP1 and to the morphing techniques (muscle-tendon attachment contours and line of action) developed in WP2.