Page 193 - Advances in Biomechanics and Tissue Regeneration

9.5 VALIDATION 189

9.5 VALIDATION

Computational models, if accurate and properly validated, are powerful and reliable tools in advancing our under-
standing of joint function in normal and perturbed conditions and in prevention and treatment programs [142–144].
Our model of the knee joint has constantly and extensively been validated at different stages of its development over
the last 20 years [11] by comparison of its predictions with available reported measurements under various loading
and boundary conditions, both passive (e.g., axial compression load [11, 36, 37, 87], F-E [84], add-abd [37, 116], A-P
drawer [81, 83, 117, 145] load-displacement conditions) and active (e.g., F-E [14, 82, 94], open-closed kinetic chain exer-
cises [126, 146, 147] and during gait [15, 20, 22, 23, 35, 36]). Overall, satisfactory agreements have been found and are
presented here for some conditions.
Results of the TF model on the contact forces/pressures/areas under axial compression forces up to 1800 N have
been compared with available measurements (Table 9.1 [37]) demonstrating good agreements. Despite differences in
femoral constraints and loadings, in vitro studies of the TF joint at full extension under axial compression [68, 70, 114,
148, 149] report similar nonlinear stiffening in the axial direction.
Under a small compression preload of 10 N, the TF FE model computed tangent add-abd angular stiffnesses of 5.38,
1.29, 1.13, and 0.83Nm/deg in frontal plane at 0-, 15-, 30-, and 45-degree joint flexion angles, respectively [37]. The
presence of compression preload (up to 1800 N) applied at MBP substantially increased foregoing TF stiffnesses at
all flexion angles [37], see Fig. 9.6. With no joint compression, Markolf et al. [161] measured comparable in vitro stiff-
ness values (mean standard deviation) of 11.0 7.5, 1.6 1.2, 1.1 0.8 and 0.8 0.9Nm/deg at 0-, 10-, 20-, and
45-degree flexion angles, respectively. Moreover, Creaby et al. [162] measured, in vivo in seated asymptomatic par-
ticipants with relaxed knee at 20-degree flexion, the midrange angular stiffness of 1.62 0.68 Nm/deg, whereas Bend-
jaballah et al. [116] and Marouane et al. [37] computed passive stiffnesses in the frontal plane of 4.5Nm/deg at full
extension and 1.46Nm/deg at 15-degree flexion, respectively.
Computed cruciate coupling and screw-home mechanism in passive TF joint during F-E were found in agreement
with the literature [84]. Screw-home motion of internal tibial rotation during joint flexion and external rotation during
extension was computed. Prediction of 16-degree internal tibial rotation at 90-degree femoral flexion fell in the range
of reported values of 14–36 [163, 164], 6.5 4 [161], 14.5 [165], 19.2 4 [166], and 20 4 degrees [167] at 90-degree flex-
ion angle. In passive unconstrained F-E of TF joint, ACL force diminished from 65 N at 10-degree hyperextension to
31 N at 90-degree flexion. This change is due to a shift in load resistance from ACL-pl bundle at hyperextension to ACL-
am bundle in flexion angles that corroborates with measurements [56]. The PCL resistance, on the other hand, initiated
at flexion angles beyond 20–30 degrees and reached 35 N (by PCL-al bundle only) at 90-degree flexion. The foregoing
predictions agree with others reporting increased force/strain in the PCL after 40–50 degrees of flexion [168] and
increased strain in ACL anteromedial bundles with flexion after 40 degrees of flexion [169].
In the passive TF joint at full extension under single and combined femoral posterior drawer and axial compression
forces, computed primary femoral laxity of 3.6 mm under 100-N posterior femoral force alone [117] compared well
with the mean anterior tibial laxities of 2.8–6.7mm reported under 100-N anterior tibial force at full extension [118,
170, 171]. The posterior translation of the femur relative to the tibia under pure compression, which is due to the pos-
terior slope of the tibia [171] agreed with reported values [170, 171]. The relative magnitude and increases with com-
pression preload of femoral translation in drawer were also in agreement with measurements [170–173]. The estimated
force of 170 N in ACL matched measured values of 75–162 N reported under 100 N drawer at full extension [174,
175]. Addition of axial compression preload further increased ACL force under drawer from 1.6 times, in agreement
with 150% in measurements [175], to more than twofold that of the applied drawer force. Prediction of the marked
increase in ACL force in the presence of compression preload [36, 37, 87] corroborates well with earlier findings
[171, 173, 176–179] but contradicts others suggesting that compression preload in drawer protects the ACL from
damage [180].
Our estimated muscle forces during the stance phase of gait [15, 20] were also compared with recorded muscle activ-
ities (EMG). Predicted activation levels in quadriceps, hamstrings, and gastrocnemii muscles were found in very sat-
isfactory agreement with values in the literature [27, 102, 181–183] and followed the same relative trends as in
measured EMG activities [134, 135]. The computed hamstring forces peaked right after the HS at 5% period [15,
20]. Reported normalized EMG activities in the superficial lateral and medial hamstrings [134] are also the highest
right at the HS and decrease thereafter. To qualitatively validate our predictions with normalized EMG measurements
collected on the same asymptomatic subjects whose kinematics and kinetics were used to drive our MS model [134],
the computed muscle forces were normalized to their maximum forces estimated at 0.6 MPa times muscle physiolog-
2
ical cross-section (Fig. 9.7). Coefficients of determination, R , are computed higher than 0.7 in all muscles. At the

I. BIOMECHANICS