(I-324) Identifying Dynamic Material Properties of Soft Tissue using Wearable Inertial Measurement Unit Sensors

Introduction:: Inertial Measurement Units (IMUs) are widely used as part of wearable sensor systems to study human motion in natural environments outside of laboratories.¹ However, wearable sensors like IMUs are often worn on top of soft tissue that gets excited during impulsive impacts, resulting in dynamic soft tissue artifact measurement errors.² Injuries such as ligament tears and mild traumatic brain injuries are associated with these impulsive impacts, thus making it difficult to measure injuries in vivo and to understand the corresponding injury mechanisms.^2,3 To account for these measurement errors, it is important we first characterize and model the errors so that we can predict and subsequently mitigate them. Since these errors are the result of soft tissue movement, modeling the underlying soft tissue properties is necessary. While there are studies that have attempted to measure the properties of human soft tissue⁴, they do so under quasi-static conditions which are not representative of dynamic impulsive impact conditions.



Our previous study empirically quantified the error caused by dynamic soft tissue artifacts to IMU sensor measurements when mild impulsive impacts were delivered to the lower extremity. Relationships between the caused error and changes in sensor mass suggested an underlying second-order model for the soft tissue. The main objective of this work is to investigate material properties of human soft tissue from analysis of the measured impacts and determine soft tissue behavior during dynamic impact situations.




Materials and Methods:: Our previous study collected mild impact data from three IMU sensors (ICM-20649) from 10 participants (consented under institutional ethics protocol REB H21-00537, 5 female, 5 male). Two IMUs were attached at the anterior and posterior centre of the dominant shank and affected by soft tissue artifacts, while the third IMU was placed on the back of the ankle and represented a ground truth reference measurement of the impulsive impacts. The ankle was fit with an ankle brace to prevent movement of the ankle joint. Impacts were generated by dropping the shank from a height of 21.6 cm onto a foam pad to reduce bouncing of the foot. Over 10 participants, we collected 1,337 impulsive impacts for analysis.

The relative vertical linear acceleration of soft tissue in each impact was obtained by computing the difference between the vertical linear acceleration of a shank IMU and the vertical linear acceleration of the ankle IMU. We then twice-integrated the relative vertical linear acceleration to obtain the relative vertical linear displacement, which represented the movement of soft tissue during impulsive impacts. Integration drift was minimized based on the assumption that the total sensor displacement from pre-impact to post-impact must be zero (the sensor and soft tissue return to their initial positions). To obtain an empirical soft tissue material behavior, we plotted the soft tissue relative vertical linear acceleration multiplied by sensor mass (5 grams) against the soft tissue relative vertical linear position. This represents a force-displacement material property curve of the underlying soft tissue.


Results, Conclusions, and Discussions:: Preliminary analysis of one impact showed a force-displacement curve mirroring common biological soft tissue material property curves with a low-strain toe region followed by a stiffer linear-elastic region (Figure 1.c). We found that the resulting force-displacement curve has a greater stiffness than those obtained from in vivo quasi-static indentation testing⁴. This indicates that soft tissue stiffens in the dynamic scenario, which has been observed in other soft tissue types such as ligaments, and is suspected to be a viscous effect.^5,6

Although we generated a force-displacement material property curve, there are a few limitations to our methodology that we are improving. First, integrating acceleration data creates drift in the subsequently estimated velocity and position. This was accounted for by applying a correction to the measured vertical linear acceleration such that the resulting twice-integration displacement was 0 mm at the end of the simulated impact. This can be further improved by also forcing the velocity post-impact to 0 m/s, as experimentally the leg was stationary for a while post-impact. Second, while we estimated a force-displacement curve by multiplying acceleration with the sensor mass, this does not account for the underlying soft tissue mass which is also part of the dynamic system. To better account for this, our experiments modulate the sensor mass to explore how soft tissue artifacts change according to the underlying material property, and that will be used to obtain a best-fit force-displacement curve against all our impact data. Despite this, the force-displacement curve based solely on the sensor mass represents a lower bound of the true force-displacement curve, and is already higher than the previously estimated quasi-static force-displacement curves of soft tissue.

Characterizing the dynamic soft tissue material properties allows us to potentially model dynamic soft tissue artifacts during impulsive impacts. This will improve our ability to predict soft tissue artifacts, in turn enabling us to mitigate the measurement errors they cause in wearable IMU systems. This can potentially assist improved research using wearable sensors in real-world settings by facilitating higher quality measurements.


Acknowledgements (Optional): :

References (Optional): : 1. Bonato P. Wearable sensors and systems. From enabling technology to clinical applications. IEEE Eng Med Biol Mag. 2010;29(3):25-36. doi:10.1109/MEMB.2010.936554

2. Wu LC, Nangia V, Bui K, et al. In Vivo Evaluation of Wearable Head Impact Sensors. Ann Biomed Eng. 2016;44(4):1234-1245. doi:10.1007/s10439-015-1423-3

3. Oh YK, Lipps DB, Ashton-Miller JA, Wojtys EM. What strains the anterior cruciate ligament during a pivot landing?. Am J Sports Med. 2012;40(3):574-583. doi:10.1177/0363546511432544

4. Pai DK, et al. The human touch: Measuring contact with real human soft tissues. ACM Transactions on Graphics. (TOG),37.4,(2018):1-12. doi:10.1145/3197517.3201296

5. Ivancic PC, Coe MP, Ndu AB, et al. Dynamic mechanical properties of intact human cervical spine ligaments. Spine J. 2007;7(6):659-665. doi:10.1016/j.spinee.2006.10.014

6. Shim VP, Liu JF & Lee VS. A Technique for Dynamic Tensile Testing of Human Cervical Spine Ligaments. Exp Mech 46, 77–89 (2006). doi:10.1007/s11340-006-5865-2