Does the Initial Guess Affect the Estimations of Knee Ligaments Properties via Optimization Procedures?

Optimization combined with experimental measurements and musculoskeletal modelling are used to estimate subject-specific ligament properties and literature values are often used for the initial guess in the process. Whether these literature values, obtained from tensile tests on cadaver specimens, affect the ligament properties estimations remains unknown. Thus, we aimed to investigate whether the initial guess affects the estimations of knee ligament properties. An experimental data set, including subject-specific anatomy, laxity tests and ligament stiffness values was obtained from one cadaver. Then, a subject specific knee computational model was developed and combined with the laxity tests provided ligament properties estimations via an optimization procedure. Two different initial guesses were used. The first one comprised of generic literature values while the second one included experimentally obtained stiffnesses. The optimization procedures with the two different initial guesses resulted in two sets of ligament properties that were different by 1.9 (1.6) kN and 2.9 (2.7)% in average for the stiffnesses and references strains. However, these sets provided models with similar laxity behavior. This is an alarming finding, and it should be investigated how such different ligament properties influence decisions and/or designs of interventions. Moreover, future research should explore experimental and computational methodologies to overcome this challenge.