DSpace https://erepo.uef.fi
Rinnakkaistallenteet Luonnontieteiden ja metsätieteiden tiedekunta
2019
Identification of locations susceptible to osteoarthritis in patients with anterior cruciate ligament reconstruction:
Combining knee joint computational
modelling with follow-up T1p and T2 imaging
Bolcos, PO
Elsevier Ltd
article
info:eu-repo/semantics/acceptedVersion
© Authors
CC BY-NC-ND https://creativecommons.org/licenses/by-nc-nd/4.0/
https://doi.org/10.1016/j.clinbiomech.2019.08.004
https://erepo.uef.fi/handle/123456789/7768
Downloaded from University of Eastern Finland's eRepository
Contents lists available atScienceDirect
Clinical Biomechanics
journal homepage:www.elsevier.com/locate/clinbiomech
Identi fi cation of locations susceptible to osteoarthritis in patients with anterior cruciate ligament reconstruction: Combining knee joint
computational modelling with follow-up T
1ρand T
2imaging
Paul O. Bolcos
a,⁎⁎, Mika E. Mononen
a, Matthew S. Tanaka
b, Mingrui Yang
c,
Juha-Sampo Suomalainen
d, Mikko J. Nissi
a,e, Juha Töyräs
a,f,g, Benjamin Ma
b, Xiaojuan Li
c, Rami K. Korhonen
a,f,⁎aDepartment of Applied Physics, University of Eastern Finland, POB 1627, FI-70211 Kuopio, Finland
bDepartment of Radiology and Biomedical Imaging, University of California San Francisco, CA-94158 San Francisco, United States of America
cDepartment of Biomedical Engineering, Cleveland Clinic, OH-44195 Cleveland, United States of America
dDepartment of Clinical Radiology, Kuopio University Hospital, POB 100, FI-70029 KUH Kuopio, Finland
eResearch Unit of Medical Imaging, Physics and Technology, University of Oulu, POB 8000, FI-90014 Oulu, Finland
fDiagnostic Imaging Centre, Kuopio University Hospital, POB 100, FI-70029 KUH Kuopio, Finland
gSchool of Information Technology and Electrical Engineering, The University of Queensland, Brisbane, QLD 4072, Australia
A R T I C L E I N F O
Keywords:
Finite-element analysis Knee joint
Articular cartilage Gait
Magnetic resonance imaging
A B S T R A C T
Background: Finite element modelling can be used to evaluate altered loading conditions and failure locations in knee joint tissues. One limitation of this modelling approach has been experimental comparison. The aims of this proof-of-concept study were: 1) identify areas susceptible to osteoarthritis progression in anterior cruciate li- gament reconstructed patients usingfinite element modelling; 2) compare the identified areas against changes in T2and T1ρvalues between 1-year and 3-year follow-up timepoints.
Methods:Two patient-specificfinite element models of knee joints with anterior cruciate ligament reconstruc- tion were created. The knee geometry was based on clinical magnetic resonance imaging and joint loading was obtained via motion capture. We evaluated biomechanical parameters linked with cartilage degeneration and compared the identified risk areas against T2and T1ρmaps.
Findings:The risk areas identified by thefinite element models matched the follow-up magnetic resonance imagingfindings. For Patient 1, excessive values of maximum principal stresses and shear strains were observed in the posterior side of the lateral tibial and femoral cartilage. For Patient 2, high values of maximum principal stresses and shear strains of cartilage were observed in the posterior side of the medial joint compartment. For both patients, increased T2and T1ρvalues between the follow-up times were observed in the same areas.
Interpretation:Finite element models with patient-specific geometries and motions and relatively simple material models of tissues were able to identify areas susceptible to post-traumatic knee osteoarthritis. We suggest that the methodology presented here may be applied in large cohort studies.
1. Introduction
The exact mechanisms behind the onset and development of os- teoarthritis (OA) are not fully understood. The incidence of OA is generally higher in patients after anterior cruciate ligament (ACL) rupture, especially with concomitant meniscal or chondral lesions (Barenius et al., 2014;Claes et al., 2013;Culvenor et al., 2015;Potter et al., 2012;Risberg et al., 2016). Additionally, a long-term follow-up
study showed little difference in OA susceptibility between conservative (exercise) or surgical treatment (ACL reconstruction, ACLR) of ruptures (Lohmander et al., 2007). Postoperatively, knee OA can be present even in short-term follow-ups of ACLR (Culvenor et al., 2015;Eckstein et al., 2015;Williams et al., 2017). One of the mechanisms leading to OA for ACLR patients could be altered joint biomechanics and excessive stresses and strains experienced by articular cartilage (Gardinier et al., 2013;Konrath et al., 2017;Wellsandt et al., 2016).
https://doi.org/10.1016/j.clinbiomech.2019.08.004 Received 27 December 2018; Accepted 7 August 2019
⁎Corresponding author.
⁎⁎Correspondence to: PO Bolcos, Department of Applied Physics, University of Eastern Finland, POB 1627, FI-70211 Kuopio, Finland E-mail addresses:paul.bolcos@uef.fi(P.O. Bolcos),rami.korhonen@uef.fi(R.K. Korhonen).
Clinical Biomechanics xxx (xxxx) xxx–xxx
0268-0033/ © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
Please cite this article as: Paul O. Bolcos, et al., Clinical Biomechanics, https://doi.org/10.1016/j.clinbiomech.2019.08.004
Traditional methods for OA diagnosis, such as Kellgren-Lawrence or Modified Fairbank systems, are unable to offer sufficient information on cartilage integrity or composition. Semi-quantitative evaluation methods, such as Whole-Organ Resonance Magnetic Score (WORMS) or Magnetic resonance imaging Osteoarthritis Knee Score (MOAKS), re- veal global longitudinal structural changes in the knee joint (Culvenor et al., 2015;Wang et al., 2016). However, both traditional methods and semi-quantitative methods are susceptible to misclassification and intra- and inter-observer variability (Ding et al., 2008;Karvonen et al., 1990;Wang et al., 2012). Measurement of T2and T1ρ(T1in rotating frame) relaxation times offer a more quantitative method to assess local changes in the knee joint (Li et al., 2014; Li and Majumdar, 2013;
Mamisch et al., 2010; Nissi et al., 2016;Rautiainen et al., 2015). In articular cartilage, T2relaxation time is mainly indicative of the col- lagen network integrity and arrangement (Hänninen et al., 2017;Nissi et al., 2004;Xia et al., 2001), while T1ρrelaxation time is primarily sensitive to proteoglycans (Duvvuri et al., 2002;Wheaton et al., 2004), but also carries sensitivity to collagen content depending on the mag- netic field strength (Blumenkrantz and Majumdar, 2007; Hänninen et al., 2017;Nieminen et al., 2017).
Imaging cannot assess altered biomechanics and excessive joint and tissue loading. Experimental and computational studies have linked collagen matrix degeneration primarily with high tensile stresses of collagenfibrils (Danso et al., 2014;Henao-Murillo et al., 2018;Hosseini et al., 2013;Kempson, 1982) and proteoglycan (PG) loss primarily with high deviatoric or shear strains of cartilage nonfibrillar matrix (Bonnevie et al., 2016;Ewers et al., 2001;Kelly and O'Connor, 1996;
Wilson et al., 2006). This has enabled the use offinite element (FE) modelling in assessing the potential biomechanical risks for the onset and progression of OA due to collagen degeneration and/or PG loss (Gardiner et al., 2016; LaValley et al., 2017;Mononen et al., 2018;
Mootanah et al., 2014). However, for better trustworthiness of the models, they should be compared against follow-up information, such
as T2and T1ρrelaxation time maps.
In a clinical setting, the computational model has to be easy to generate and time for the converged solution has to be short. Some of the above-mentioned models applied complex materials, such asfibril- reinforced poro(visco)elastic, to describe articular cartilage, meniscus and ligaments (Dabiri and Li, 2015;Gu and Li, 2011;Mononen et al., 2018;Mootanah et al., 2014). This is simultaneously time consuming (in terms of implementation) and computationally demanding. Fur- thermore, for models that include all major knee joint structures with muscle forces, generation and simulation times become even longer (Bolcos et al., 2018;Orozco et al., 2018). Recent studies have shown that simpler FE models, in terms of geometry and motion (Bolcos et al., 2018), cartilage material properties (Klets et al., 2016) and ligament formulation (Orozco et al., 2018), produce similar results with more complex models. This enables a reduction in FE model generation and computation times. This kind of FE models for joint mechanics have not been generated before for ACLR patients with patient-specific motion.
Further, to our knowledge, patient-specific OA predictions from the FE models have not been compared earlier against experimentally eval- uated local changes in the knee joint cartilage, as determined by changes in T2and T1ρrelaxation times between follow-up timepoints.
The objectives of this proof of concept study were two-fold: (1) Using a relatively fast FE modelling approach, with a proven ability to capture mechanical responses of cartilage, to evaluate knee cartilage mechanics in patients with ACLR at 1-year follow-up, and identify areas susceptible to OA progression, due to collagen damage and/or PG loss;
(2) Compare the identified areas for collagen degeneration and PG depletion against local changes in T2and T1ρrelaxation times between the 1-year and 3-year follow-up timepoints. Our hypothesis was that the onset and development of OA in ACLR patients is patient-specific and the areas susceptible to OA at 1-year timepoint can be identified by using FE modelling and matched with local changes in T2and T1ρre- laxation times of cartilage.
Fig. 1.Workflow of the study. a) Knee joint MR image segmentation; b) Knee joint rotations and ground reaction forces from motion capture; c) FE model overview, with geometry from a) and motion from b); d) Maximum principal stress distribution on the tibial cartilage. Areas susceptible to OA are indicated in black; e) T2and T1ρmaps used for verifying the progression of OA.
2. Methods
The workflow of the study is shown inFig. 1. This study includes two patient-specific FE models of two subjects with ACLR. Information on both the knee joint geometry and motion was incorporated from manually segmented high-resolution 3D MR images (Fig. 1a) and mo- tion capture gait data (Fig. 1b), respectively. The included soft-tissues were femoral and tibial cartilages and menisci, with collateral (MCL &
LCL) and cruciate (ACL & PCL) ligaments (Fig. 1c). The FE model results (Fig. 1d) were then compared against follow-up information: T2and T1ρ
maps (Fig. 1e).
2.1. Data acquisition
The magnetic resonance (MR) image acquisition and motion capture were performed at the University of California, San Francisco (UCSF).
Both subjects gave informed consent and data acquisition was approved by and carried out in accordance with the rules and regulations of the Institutional Review Board under the Human Research Protection Program at UCSF. For each patient, two MR sequences were acquired at 1-year and 3-year follow-up timepoints after the ACLR surgery.
Additionally, at each follow-up timepoint the subject gait data was measured using a previously established protocol (Samaan et al., 2017).
The 1-year timepoint was used to predict the location susceptible to OA using FE modelling. The 3-year timepoint was used to verify the pro- gression of OA predicted from the 1-year timepoint. Details on the MRI acquisition and motion capture are given in Supplementary Materials.
2.2. FE model construction
MRI and motion capture data were transferred to the University of Eastern Finland (UEF), where computational models were generated.
There is a data transfer agreement between UCSF and UEF. The methodology used to generate the FE models was identical to a previous study (Bolcos et al., 2018), and is summarized in Supplementary ma- terials. In that study, it was shown that simpler knee models can pro- duce similar cartilage responses with more complex models. The FE model with motion implemented using kinetics and kinematics (forces and rotations), and without patella and quadriceps forces, produced reaction forces and contact pressures within physiological limits (Gilbert et al., 2014;Konrath et al., 2017;Kutzner et al., 2010;Pizzolato et al., 2017). Details of the material properties for each soft tissue are shown inTable 1. Simpler models are desired when the purpose is to- wards clinical implementation. In this study, this relatively simple ap- proach with kinetic-kinematic motion implementation was used.
2.3. Analysis
2.3.1. T2and T1ρmaps
Tibial and femoral cartilage were manually segmented from the
combined multi-slice sequence at both 1-year at 3-year follow-up timepoints. The relaxation times were calculated using a two-para- metric mono-exponentialfit with Aedes (Niskanen, 2006) and in-house written plugins for Matlab.
2.3.2. FE analysis
Based on the experimentalfindings (Danso et al., 2014;Kempson, 1982; Mononen et al., 2016) and previous computer simulations (Mononen et al., 2016), maximum principal stresses above 7 MPa were assumed to trigger collagen network degeneration. Generally, experi- mental findings (D'lima et al., 2001; Ewers et al., 2001; Kelly and O'Connor, 1996; Loening et al., 2000;Wilson et al., 2006;Zamli and Sharif, 2011) and previous computer simulations (Hosseini et al., 2014;
Mononen et al., 2018) link tissue strains above 30% with condrocyte apoptosis and subsequent PG loss. Here, shear strains above 32% were assumed to lead to PG loss (Bonnevie et al., 2016;Hashimoto et al., 2009;Li et al., 2013). Here we evaluated:
1.σtensileandγabsas a function of stance. The peak values of maximum principal stress (σtensile) and absolute shear strain (γabs) were calcu- lated on the tibiofemoral contact area (cartilage-cartilage contact area) as a function of stance.
2.σtensileandγabsdistribution maps. To identify the locations prone to collagen network damage and/or PG loss, theσtensileandγabs dis- tributions were calculated for each compartment (i.e. medial or lateral tibial/femoral cartilage). We evaluated the peak values of maximum principal stresses and absolute shear strains for each element.
2.4. Comparison of the FE model and MRI
The FE model results were verified against changes in T2and T1ρ relaxation times. These are among the most established quantitative MRI parameters for articular cartilage and were shown to be highly sensitive to collagen and PG content (Duvvuri et al., 2002;Hänninen et al., 2017; Li et al., 2007; Nissi et al., 2004; Wang et al., 2016;
Wheaton et al., 2004;Xia et al., 2001).
2.4.1. Total volumes
To compare the identified risk areas with follow-up information the following steps were needed:
1. For theσtensileandγabsdistribution maps, we defined volumes-of- interest (VOI) for each compartment. The VOI was defined as the total volume in which the respective thresholds were exceeded. If neitherσtensilenorγabsexceeded the thresholds, the VOI was defined as“0”.
2. For each VOI from step 1, we calculated the total volume of the VOI as a percentage of the total volume of each compartment, reflecting the percentage from the total volume susceptible to damage.
Table 1
Material parameters of cartilage, meniscus and ligaments used in the FE models.
Transversely isotropic (poro)elastic Ep(MPa) Et(MPa) νp(−) νtp(−) Gt(MPa) k (10−15m4/Ns) e0(−)
Cartilage (Klets et al., 2016) 24 0.46 0.42 0.06 12 1 4
Meniscus (Danso et al., 2014;Klets et al., 2016;Vaziri et al., 2008;Wilson et al., 2004) 20 159.6 0.30 0.01 50 – –
Bi-linear springs Stiffness (N/mm) Pre-strain (%)
ACL (Gantoi et al., 2013;Haut Donahue et al., 2002) 380 5
PCL (Gantoi et al., 2013;Momersteeg et al., 1995) 200 5
MCL(Gantoi et al., 2013;Momersteeg et al., 1995) 100 4
LCL (Gantoi et al., 2013;Momersteeg et al., 1995) 100 4
Parameters:Ep–in-plane Young's modulus,Et–out-of-plane Young's modulus,νp–in-plane Poisson's ratio,νtp–out-of-plane Poisson's ratio,Gt–out-of-plane shear modulus,k–permeability,e0–initial void ratio.
P.O. Bolcos, et al. Clinical Biomechanics xxx (xxxx) xxx–xxx
3
3. For the T2 and T1ρmaps, from both 1-year and 3-year follow-up timepoints we defined VOIs for each compartment. The VOI was defined as the volume of cartilage with either T2or T1ρrelaxation times above 60 ms. This value is above the literature reported value of 50 ms for healthy cartilage (Bolbos et al., 2008;Li et al., 2011;
Stahl et al., 2009;Surowiec et al., 2014;Van Rossom et al., 2017).
Similarly, if no relaxation time exceeded this limit, the VOI was defined as“0”.
4. For each VOI from step 3, we calculated the volume of the VOI as a percentage from the total volume of the compartment at both 1-year and 3-year follow-up timepoints.
5. From step 4, we subtracted the VOI at 3-year timepoint from the VOI at 1-year timepoint, reflecting the percentage of potentially da- maged tissue from the total volume of each compartment.
Thus, we could evaluate changes in the relaxation times between 1- year and 3-year follow-up times and compare them with the areas susceptible to degeneration as predicted by the FE model. An example of steps 1–4 is shown in Supplementary materials.
2.4.2. Sagittal slices
Due to the slice thickness of 4 mm of both T2and T1ρMR images, we could not ensure the accuracy of the total volume calculated in steps 4 and 5. Therefore, a slice-by-slice comparison between the relaxation times and FE models was required. Since in the sagittal plane the re- solution of the MR image was the best, sagittal slices from theσtensile, γabs, T2and T1ρmaps were taken as follows:
1. For T2and T1ρmaps, the sagittal slice was located in the center of the previously defined VOI. The location of this slice was approxi- mated by calculating the number of slices to the edge of lateral side and multiplying with the slice thickness (4 mm).
2. For theσtensileandγabsdistribution maps, the sagittal slice was ac- quired from the same location as in step 1 (Fig. 2a and b).
3. Results
3.1. Stance
The FE model revealed that on the lateral tibial cartilage in Patient
1, σtensile exceeded the 7 MPa threshold for collagen degeneration
through the entire stance phase (Fig. 3a), whileγabsvalues were above the threshold of 32% for PG loss between 20% and 80% of the stance phase (Fig. 3b). On the lateral femoral cartilage, the thresholds were exceeded for bothσtensileandγabsafter the second peak force (80% of the stance) (Fig. 3c,d). On both medial tibial and femoral cartilages, neither σtensilenorγabsexceeded the degeneration thresholds (Fig. 3a–d).
For Patient 2, σtensile exceeded the threshold through the entire stance phase on the medial tibial cartilage (Fig. 3e), whileγabsvalues were above the threshold for degeneration at 0–20% and 50–80% (after midstance) of the stance phase (Fig. 3f). On the lateral tibial cartilage, neitherσtensilenorγabsexceeded the degeneration thresholds (Fig. 3e,f).
For the femoral cartilage, the thresholds were exceeded for bothσtensile
andγabsat 30–50% of the stance phase in the lateral and at ~20% in the medial joint compartments (Fig. 3g,h).
3.2. Distribution
For Patient 1,σtensileexceeded the threshold of 7 MPa on the pos- terior side of both the lateral tibial and femoral cartilage (Fig. 4a and b).
Theγabsvalues also exceeded the threshold of 32% with a similar dis- tribution asσtensile(not shown). On the medial tibial and femoral car- tilage, neitherσtensilenorγabsexceeded the thresholds.
For Patient 2,σtensileexceeded the threshold for collagen damage on the posterior side of the medial tibial and femoral cartilage (Fig. 4c and d). Further, in the lateral joint compartment, theσtensilevalues exceeded the threshold in the center of the cartilage, but the area with high va- lues was smaller than that in the medial joint compartment. Theγabs
values were only slightly over the threshold of proteoglycan loss (PG) if at all.
3.3. Volume change
The total volume susceptible to OA identified by the FE model matched adequately with the total volume of increased T2and T1ρre- laxation times (Fig. 5). For Patient 1, the FE model revealed that in
~14% of the lateral tibial cartilage and ~7% of the lateral femoral cartilage volume, theσtensilevalues were above the threshold. Theγabs
values were above the chosen threshold in ~6% of the lateral tibial cartilage and ~7% of the lateral femoral cartilage volume (Fig. 5a).
Similarly, ~16% of the lateral tibial cartilage and ~6% of the lateral femoral cartilage volume experienced increased values for both T2and T1ρduring the follow-up times. In the medial joint compartment, nei- ther σtensilenorγabs exceeded the thresholds. Similarly, in the medial tibial cartilage, neither T2nor T1ρwere changed between the follow-up times, while in the femoral side they were increased in ~3% of the total cartilage volume (Fig. 5a).
For Patient 2, the FE model revealed highσtensilevalues in ~2% of the lateral compartment and ~3% of the medial compartment cartilage volume. Highγabs values were only seen at maximum of ~1% of the lateral femoral cartilage volume (Fig. 5b). Similarly, ~2.5% of the lateral tibial cartilage and ~1.5% of the lateral femoral cartilage vo- lume experienced increased values for both T2 and T1ρ during the follow-up times. In the medial tibial cartilage, both T2and T1ρvalues
Fig. 2.a) and b) Sagittal slice locations for the FE models. Note that slice thicknesses for both the FE models and T2/T1ρmaps are indicated on the right.
increased between the follow-up timepoints in ~7.5% of the total vo- lume, while in the femoral side they increased in ~1.5% of the cartilage volume (Fig. 5b).
3.4. Sagittal slices
For Patient 1, the FE model results and T2and T1ρmaps of both the lateral (Fig. 6a) and medial (Fig. 6b) compartments showed an ade- quate correspondence. In the areas of the lateral compartment with highσtensileorγabsvalues (dark gray areas inFig. 6a), the T2and T1ρ
values were also elevated. The relaxation times more than doubled in the same areas where excessiveσtensileand/orγabsvalues were seen. In the medial femoral compartment, only local increases in T2 and T1ρ
relaxation times between the follow-up timepoints were observed, while the medial tibial cartilage was unaffected. The FE model showed neither highσtensilenorγabsvalues, above the chosen threshold, for ei- ther medial femoral or tibial cartilage.
For Patient 2, the FE model showed only slightly elevatedσtensileand γabsvalues near the cartilage surface in the lateral joint compartment (Fig. 7a), while the T2and T1ρvalues were close to the literature re- ported values for healthy cartilage. In the medial joint compartment (Fig. 7b), highσtensileandγabsvalues were seen on the posterior side of the medial tibial and femoral cartilage. For the T2and T1ρrelaxation times of the medial joint compartment, slightly elevated values during the follow-up were seen throughout the contacting surfaces.
4. Discussion
In the present proof-of-concept study, two FE models of patients with ACL reconstruction were created. The knee joint geometries were based on manually segmented MRI images and the knee joint motions were based on motion capture. Each model included tibial cartilage, femoral cartilage and menisci with collateral and cruciate ligaments. To reduce model complexity and calculation times, a transversely isotropic poroelastic material was used for cartilage and a transversely isotropic
elastic material for menisci. The analysis was divided in two parts. First, we identified potential failure areas for both tibial and femoral cartilage using FE modelling. Then, we compared these areas against quantita- tive follow-up T2and T1ρrelaxation times. The potential failure areas predicted by the FE model matched adequately with the follow-up MRI information for both patients. Our results suggest that a relatively simple FE model, in terms of geometry, motion and materials, has po- tential to identify areas susceptible to cartilage degeneration and may be applied in a fast evaluation of subjects with traumatic ligament in- juries and reconstructions.
4.1. Patient 1
Based on the FE model results, possible collagen damage and de- generation throughσtensile was predicted to occur in both the lateral tibial (14% of the cartilage volume) and lateral femoral cartilage (7% of the cartilage volume). In agreement with the simulation results, pri- marily collagen-sensitive T2(Nissi et al., 2004;Xia et al., 2001) more than doubled in both the lateral tibial (16% of cartilage volume) and lateral femoral cartilage (5% of cartilage volume). Moreover, the FE model revealed that the posterior side of the lateral joint compartment was the area most susceptible to collagen damage. This result was supported by the elevated T2during the follow-up. In the lateral fe- moral cartilage, highγabsvalues indicated PG loss in ~7% of the vo- lume. This result was confirmed by the elevated T1ρduring the follow- up, also in ~5% of the volume. This parameter has been considered to be mostly sensitive to PGs (Duvvuri et al., 2002). Despite high values of γabs in the lateral tibial cartilage only in ~6% of the total cartilage volume, indicating PG loss, the PG-sensitive T1ρ relaxation time was elevated in ~16% of the lateral tibial cartilage volume.
In the medial tibial cartilage, neither the FE model nor the experi- mental follow-up information indicated any degenerative signs by the collagen-specific (σtensileand T2) and PG-specific (γabs and T1ρ) para- meters. However, in the medial femoral cartilage, despite elevated T2
and T1ρ values in local areas (~4% of the total volume) during the Fig. 3.Maximum values of maximum principal stresses and absolute shear strains as a function of stance for Patient 1 in the tibial (a and b) and femoral cartilage (c and d). Maximum values of maximum principal stresses and absolute shear strains as a function of stance for Patient 2 for the tibial (e and f) and femoral cartilage (g and h). The values are calculated from the cartilage-cartilage contact area in both the tibial and femoral cartilage. Approximated thresholds for collagen degeneration and PG loss are indicated with dashed lines.
P.O. Bolcos, et al. Clinical Biomechanics xxx (xxxx) xxx–xxx
5
follow-up, the FE model did not predict tensile stresses or shear strains above the chosen thresholds. Taking into consideration that no changes were seen in the tibial compartment in any of the parameters, this in- crease in the femoral side may be caused by factors that could not be considered in the model. See more from limitations below.
4.2. Patient 2
Based on the FE model results, possible collagen damage and de- generation throughσtensilewas predicted to occur in the medial tibial (3% of cartilage volume) and femoral cartilage (3% of cartilage Fig. 4.Axial views of maximum principal (tensile) stress distributions on the tibial and femoral cartilage for Patient 1 (a and b, respectively) and Patient 2 (c and d, respectively). Peak values for tensile stresses are also indicated.
Fig. 5.Degenerated volumes predicted by the FE model (σtensileandγabs) and measured from the MRI follow-up (T2and T1ρ) as a percentage of the entire cartilage volume. The volumes were calculated for the lateral and femoral tibial and femoral cartilage of Patient 1 (a) and Patient 2 (b).
Fig. 6.Sagittal slices for the FE model and corresponding sagittal T2and T1ρmap slices at both 1-year and 3-year follow-up timepoints for the lateral and medial compartments of Patient 1 (a and b). Slice locations are indicated inFig. 2a and arrows indicate the peak values. Note: All values above the selected degeneration thresholds in the FE models (7 MPa for tensile stress and 32% for shear strain) are shown in dark gray. T2and T1ρrelaxation times above 100 ms are shown in dark red. (For interpretation of the references to colour in thisfigure legend, the reader is referred to the web version of this article.)
Fig. 7.Sagittal slices for the FE model and corresponding sagittal T2and T1ρmap slices at both 1-year and 3-year follow-up timepoints for the lateral and medial compartments of Patient 2 (a and b). Slice locations are indicated inFig. 2b and arrows indicate the peak values. Note: All values above the selected degeneration threshold (7 MPa for tensile stress and 32% for shear strain for the FE model) are shown in dark gray. T2and T1ρrelaxation times above 100 ms are shown in dark red.
(For interpretation of the references to colour in thisfigure legend, the reader is referred to the web version of this article.)
P.O. Bolcos, et al. Clinical Biomechanics xxx (xxxx) xxx–xxx
7
volume). In agreement with the simulation results, the collagen-sensi- tive T2was elevated in the same sites during the follow-up, greatest changes seen in the medial tibial cartilage. Moreover, the FE model predicted that the posterior side of the medial joint compartment is susceptible to collagen damage, which was supported by the T2ana- lysis. In the medial femoral cartilage, bothγabsand T1ρindicated PG loss only in ~1% of the cartilage volume. On the other hand, in the medial tibial cartilage, the PG-sensitive T1ρrelaxation time suggested altera- tions in a much larger area than what was predicted by theγabs. In both the lateral tibial and femoral cartilage, both the FE model and the ex- perimental MRI follow-up information indicated only minor degen- erative signs at the posterior side of the joint and primarily at the contact area.
4.3. Limitations
This study has a few limitations, expanded upon in Supplementary Materials. One limitation is that the mechanical properties of cartilage were not patient-specific. However, the sensitivity study in Supplementary Materials indicated that, despite different material parameters of cartilage, all models identified the same locations sus- ceptible to cartilage degeneration. Identification of locations likely to degenerate may reveal which kind of rehabilitation exercises would be the most beneficial for the patient to minimize stress and strain con- centrations in locations at the highest risk for the progression of OA (Pizzolato et al., 2017).
The thresholds for defining degenerated volumes from MRI (re- laxation time > 60 ms) are not unbiased, as the values depend on the specific implementation of the respective measurements (Matzat et al., 2015). Furthermore, both T2and T1ρhave demonstrated sensitivity to the orientation of the tissue in the magneticfield, complicating spatial analysis (Hänninen et al., 2017). However, in this longitudinal study, the same measurement protocol, system and analysis was used at both follow-up times, alleviating issues related to potential differences in the results.
We also acknowledge that the sensitivity of the mechanical and MRI parameters to either collagen degeneration or PG loss is not un- ambiguous. With only two subjects, it is difficult to correlate the FE model predictions and MRI follow-up information. Studies with higher number of patients are needed. This should allow for a comprehensive statistical analysis and help tune the threshold levels. In conjunction with the limitations presented here and in Supplementary Materials, these factors may account for some of the discrepancies between the FE model results and the follow-up information. However, this is a proof- of-concept study showing that it is possible to predict potential cartilage degeneration areas using patient-specific FE models.
4.4. Clinical application
Generation of a subject-specific computational model requires a lot of manual work and time in segmentation of soft tissues, meshing and making models to converge. In future studies, the methodology pre- sented here should be coupled with semi-automatic or fully automatic segmentation techniques (Chandra et al., 2016; Dodin et al., 2010;
Folkesson et al., 2007;Lee et al., 2014;Liukkonen et al., 2017b;Paproki et al., 2014; Shan et al., 2014;Tamez-Pena et al., 2012;Yang et al., 2015;Yu et al., 2016) and with automated meshing tools (Rodriguez- Vila et al., 2017). As motion capture systems are not readily available in clinical settings, a simple and fast method should be developed to ob- tain and implement patient's gait. For instance, differences between patient-specific and population-specific (e.g. normal, early/advanced/
medial OA populations) motions could be studied. If the population- specific approach would produce similar results with the patient-spe- cific method, it could be used without motion capture. These afore- mentioned methods would ease the applications of the FE models in large cohort studies to identify areas susceptible to OA development
(Gardiner et al., 2016;Hosseini et al., 2014;Kar et al., 2016;Liukkonen et al., 2017a). They could also be coupled with iterative computational models to predict collagen degeneration and proteoglycan loss for a patient in time (Gardiner et al., 2016;LaValley et al., 2017;Mononen et al., 2018).
In conclusion, our results suggest that the FE models, as presented here, could be used to identify areas susceptible to OA onset and de- velopment. They would be particularly useful in assessing the effect of surgical interventions, such as ACL reconstruction. Moreover, it would be possible to evaluate non-surgical management options for avoiding or delaying OA onset and/or progression, such as the gait retraining method proposed by (Pizzolato et al., 2017). In conjunction with the improvements mentioned above, the presented methodology could provide a pathway towards clinical implementation.
Declaration of competing interest
The authors have no potential conflicts of interest to declare.
Acknowledgements
This project has received funding from the Doctoral Programme in Science, Technology and Computing (SCITECO) of the University of Eastern Finland, the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement No 755037), Academy of Finland (grants 285909, 307932 and 286526), Sigrid Jusélius Foundation, and National Institutes of Health (NIH/NIAMS P50 AR060752). CSC-IT Center for Science, Finland, is acknowledged for providing computing resources.
Appendix A. Supplementary data
Supplementary data to this article can be found online athttps://
doi.org/10.1016/j.clinbiomech.2019.08.004.
References
Barenius, B., Ponzer, S., Shalabi, A., Bujak, R., Norlén, L., Eriksson, K., 2014. Increased risk of osteoarthritis after anterior cruciate ligament reconstruction: a 14-year follow- up study of a randomized controlled trial. Am. J. Sports Med. 42, 1049–1057.https://
doi.org/10.1177/0363546514526139.
Blumenkrantz, G., Majumdar, S., 2007. Quantitative magnetic resonance imaging of ar- ticular cartilage in osteoarthritis. Eur. Cell. Mater. 13, 76–86.
Bolbos, R.I., Ma, C.B., Link, T.M., Majumdar, S., Li, X., 2008. In vivo T1ρquantitative assessment of knee cartilage after anterior cruciate ligament injury using 3 tesla magnetic resonance imaging. Investig. Radiol. 43, 782–788.https://doi.org/10.
1097/RLI.0b013e318184a451.
Bolcos, P.O., Mononen, M.E., Mohammadi, A., Ebrahimi, M., Tanaka, M.S., Samaan, M.A., Souza, R.B., Li, X., Suomalainen, J.-S., Jurvelin, J.S., Töyräs, J., Korhonen, R.K., 2018. Comparison between kinetic and kinetic-kinematic driven knee jointfinite element models. Sci. Rep. 8, 17351.https://doi.org/10.1038/s41598-018-35628-5.
Bonnevie, E.D., Delco, M.L., Jasty, N., Bartell, L., Fortier, L.A., Cohen, I., Bonassar, L.J., 2016. Chondrocyte death and mitochondrial dysfunction are mediated by cartilage friction and shear strain. Osteoarthr. Cartil. 24, S46.https://doi.org/10.1016/j.joca.
2016.01.107.
Chandra, S.S., Surowiec, R., Ho, C., Xia, Y., Engstrom, C., Crozier, S., Fripp, J., 2016.
Automated analysis of hip joint cartilage combining MR T2 and three-dimensional fast-spin-echo images. Magn. Reson. Med. 75, 403–413.https://doi.org/10.1002/
mrm.25598.
Claes, S., Hermie, L., Verdonk, R., Bellemans, J., Verdonk, P., 2013. Is osteoarthritis an inevitable consequence of anterior cruciate ligament reconstruction? A meta-analysis.
Knee Surgery, Sport. Traumatol. Arthrosc. 21, 1967–1976.https://doi.org/10.1007/
s00167-012-2251-8.
Culvenor, A.G., Collins, N.J., Guermazi, A., Cook, J.L., Vicenzino, B., Khan, K.M., Beck, N., van Leeuwen, J., Crossley, K.M., 2015. Early knee osteoarthritis is evident one year following anterior cruciate ligament reconstruction: a magnetic resonance imaging evaluation. Arthritis Rheumatol 67, 946–955.https://doi.org/10.1002/art.
39005.
Dabiri, Y., Li, L., 2015. Focal cartilage defect compromisesfluid-pressure dependent load support in the knee joint. Int. J. Numer. Method. Biomed. Eng. 31, e02713.https://
doi.org/10.1002/cnm.2713.
Danso, E.K., Honkanen, J.T.J., Saarakkala, S., Korhonen, R.K., 2014. Comparison of nonlinear mechanical properties of bovine articular cartilage and meniscus. J.
Biomech. 47, 200–206.https://doi.org/10.1016/j.jbiomech.2013.09.015.
Ding, C., Parameswaran, V., Cicuttini, F., Burgess, J., Zhai, G., Quinn, S., Jones, G., 2008.
Association between leptin, body composition, sex and knee cartilage morphology in older adults: the Tasmanian older adult cohort (TASOAC) study. Ann. Rheum. Dis.
67, 1256–1261.https://doi.org/10.1136/ard.2007.082651.
D'lima, D.D., Hashimoto, S., Chen, P.C., Colwell, C.W., Lotz, M.K., 2001. Human chon- drocyte apoptosis in response to mechanical injury. Osteoarthr. Cartil. 9, 712–719.
https://doi.org/10.1053/joca.2001.0468.
Dodin, P., Pelletier, J., Martel-Pelletier, J., Abram, F., 2010. Automatic human knee cartilage segmentation from 3-D magnetic resonance images. IEEE Trans. Biomed.
Eng. 57, 2699–2711.https://doi.org/10.1109/TBME.2010.2058112.
Duvvuri, U., Kudchodkar, S., Reddy, R., Leigh, J.S., 2002. T1ρrelaxation can assess longitudinal proteoglycan loss from articular cartilage in vitro. Osteoarthr. Cartil. 10, 838–844.https://doi.org/10.1053/joca.2002.0826.
Eckstein, F., Wirth, W., Lohmander, L.S., Hudelmaier, M.I., Frobell, R.B., 2015. Five-year followup of knee joint cartilage thickness changes after acute rupture of the anterior cruciate ligament. Arthritis Rheumatol 67, 152–161.https://doi.org/10.1002/art.
38881.
Ewers, B.J., Dvoracek-Driksna, D., Orth, M.W., Haut, R.C., 2001. The extent of matrix damage and chondrocyte death in mechanically traumatized articular cartilage ex- plants depends on rate of loading. J. Orthop. Res. 19, 779–784.https://doi.org/10.
1016/S0736-0266(01)00006-7.
Folkesson, J., Dam, E.B., Olsen, O.F., Pettersen, P.C., Christiansen, C., 2007. Segmenting articular cartilage automatically using a voxel classification approach. IEEE Trans.
Med. Imaging 26, 106–115.https://doi.org/10.1109/TMI.2006.886808.
Gantoi, F.M., Brown, M.A., Shabana, A.A., 2013. Finite element modeling of the contact geometry and deformation in biomechanics applications 1. J. Comput. Nonlinear Dyn. 8, 041013.https://doi.org/10.1115/1.4024541.
Gardiner, B.S., Woodhouse, F.G., Besier, T.F., Grodzinsky, A.J., Lloyd, D.G., Zhang, L., Smith, D.W., 2016. Predicting knee osteoarthritis. Ann. Biomed. Eng. 44, 222–233.
https://doi.org/10.1007/s10439-015-1393-5.
Gardinier, E.S., Manal, K., Buchanan, T.S., Snyder-Mackler, L., 2013. Altered loading in the injured knee after ACL rupture. J. Orthop. Res. 31, 458–464.https://doi.org/10.
1002/jor.22249.
Gilbert, S., Chen, T., Hutchinson, I.D., Choi, D., Voigt, C., Warren, R.F., Maher, S.A., 2014.
Dynamic contact mechanics on the tibial plateau of the human knee during activities of daily living. J. Biomech. 47, 2006–2012.https://doi.org/10.1016/j.jbiomech.
2013.11.003.
Gu, K.B., Li, L.P., 2011. A human knee joint model consideringfluid pressure andfiber orientation in cartilages and menisci. Med. Eng. Phys. 33, 497–503.https://doi.org/
10.1016/j.medengphy.2010.12.001.
Hänninen, N., Rautiainen, J., Rieppo, L., Saarakkala, S., Nissi, M.J., 2017. Orientation anisotropy of quantitative MRI relaxation parameters in ordered tissue. Sci. Rep. 7, 9606.https://doi.org/10.1038/s41598-017-10053-2.
Hashimoto, S., Nishiyama, T., Hayashi, S., Fujishiro, T., Takebe, K., Kanzaki, N., Kuroda, R., Kurosaka, M., 2009. Role of p53 in human chondrocyte apoptosis in response to shear strain. Arthritis Rheum. 60, 2340–2349.https://doi.org/10.1002/art.24706.
Haut Donahue, T.L., Howell, S.M., Hull, M.L., Gregersen, C., 2002. A biomechanical evaluation of anterior and posterior tibialis tendons as suitable single-loop anterior cruciate ligament grafts. Arthrosc. J. Arthrosc. Relat. Surg. 18, 589–597.https://doi.
org/10.1053/jars.2002.32617.
Henao-Murillo, L., Ito, K., van Donkelaar, C.C., 2018. Collagen damage location in ar- ticular cartilage differs if damage is caused by excessive loading magnitude or rate.
Ann. Biomed. Eng. 46, 605–615.https://doi.org/10.1007/s10439-018-1986-x.
Hosseini, S.M., Veldink, M.B., Ito, K., van Donkelaar, C.C., 2013. Is collagenfiber damage the cause of early softening in articular cartilage? Osteoarthr. Cartil. 21, 136–143.
https://doi.org/10.1016/j.joca.2012.09.002.
Hosseini, S.M., Wilson, W., Ito, K., van Donkelaar, C.C., 2014. A numerical model to study mechanically induced initiation and progression of damage in articular cartilage.
Osteoarthr. Cartil. 22, 95–103.https://doi.org/10.1016/j.joca.2013.10.010.
Kar, S., Smith, D.W., Gardiner, B.S., Li, Y., Wang, Y., Grodzinsky, A.J., 2016. Modeling IL- 1 induced degradation of articular cartilage. Arch. Biochem. Biophys. 594, 37–53.
https://doi.org/10.1016/j.abb.2016.02.008.
Karvonen, R.L., Negendank, W.G., Fraser, S.M., Mayes, M.D., An, T., Fernandez-Madrid, F., 1990. Articular cartilage defects of the knee: correlation between magnetic re- sonance imaging and gross pathology. Ann. Rheum. Dis. 49, 672–675.https://doi.
org/10.1136/ard.49.9.672.
Kelly, P.A., O'Connor, J.J., 1996. Transmission of rapidly applied loads through articular cartilage. Part 2: cracked cartilage. Proc. Inst. Mech. Eng. H. 210, 39–49.https://doi.
org/10.1243/PIME_PROC_1996_210_389_02.
Kempson, G.E., 1982. Relationship between the tensile properties of articular cartilage from the human knee and age. Ann. Rheum. Dis. 41, 508–511.https://doi.org/10.
1136/ard.41.5.508.
Klets, O., Mononen, M.E., Tanska, P., Nieminen, M.T., Korhonen, R.K., Saarakkala, S., 2016. Comparison of different material models of articular cartilage in 3D compu- tational modeling of the knee: data from the osteoarthritis initiative (OAI). J.
Biomech. 49, 3891–3900.https://doi.org/10.1016/j.jbiomech.2016.10.025.
Konrath, J.M., Saxby, D.J., Killen, B.A., Pizzolato, C., Vertullo, C.J., Barrett, R.S., Lloyd, D.G., 2017. Muscle contributions to medial tibiofemoral compartment contact loading following ACL reconstruction using semitendinosus and gracilis tendon grafts. PLoS One 12, e0176016.https://doi.org/10.1371/journal.pone.0176016.
Kutzner, I., Heinlein, B., Graichen, F., Bender, A., Rohlmann, A., Halder, A., Beier, A., Bergmann, G., 2010. Loading of the knee joint during activities of daily living measured in vivo infive subjects. J. Biomech. 43, 2164–2173.https://doi.org/10.
1016/j.jbiomech.2010.03.046.
LaValley, M.P., Lo, G.H., Price, L.L., Driban, J.B., Eaton, C.B., McAlindon, T.E., 2017.
Development of a clinical prediction algorithm for knee osteoarthritis structural
progression in a cohort study: value of adding measurement of subchondral bone density. Arthritis Res. Ther. 19, 95.https://doi.org/10.1186/s13075-017-1291-3.
Lee, J.-G., Gumus, S., Moon, C.H., Kwoh, C.K., Bae, K.T., 2014. Fully automated seg- mentation of cartilage from the MR images of knee using a multi-atlas and local structural analysis method. Med. Phys. 41, 092303.https://doi.org/10.1118/1.
4893533.
Li, X., Majumdar, S., 2013. Quantitative MRI of articular cartilage and its clinical ap- plications. J. Magn. Reson. Imaging 38, 991–1008.https://doi.org/10.1002/jmri.
24313.
Li, X., Benjamin Ma, C., Link, T.M., Castillo, D.-D., Blumenkrantz, G., Lozano, J., Carballido-Gamio, J., Ries, M., Majumdar, S., 2007. In vivo T1ρand T2 mapping of articular cartilage in osteoarthritis of the knee using 3T MRI. Osteoarthr. Cartil. 15, 789–797.https://doi.org/10.1016/j.joca.2007.01.011.
Li, X., Kuo, D., Theologis, A., Carballido-Gamio, J., Stehling, C., Link, T.M., Ma, C.B., Majumdar, S., 2011. Cartilage in anterior cruciate ligament-reconstructed knees: MR imaging T1{rho} and T2–initial experience with 1-year follow-up. Radiology 258, 505–514.https://doi.org/10.1148/radiol.10101006.
Li, Y., Frank, E.H., Wang, Y., Chubinskaya, S., Huang, H.-H., Grodzinsky, A.J., 2013.
Moderate dynamic compression inhibits pro-catabolic response of cartilage to me- chanical injury, tumor necrosis factor-αand interleukin-6, but accentuates de- gradation above a strain threshold. Osteoarthr. Cartil. 21, 1933–1941.https://doi.
org/10.1016/j.joca.2013.08.021.
Li, X., Wyatt, C., Rivoire, J., Han, E., Chen, W., Schooler, J., Liang, F., Shet, K., Souza, R., Majumdar, S., 2014. Simultaneous acquisition of T 1ρand T 2 quantification in knee cartilage: repeatability and diurnal variation. J. Magn. Reson. Imaging 39, 1287–1293.https://doi.org/10.1002/jmri.24253.
Liukkonen, M.K., Mononen, M.E., Klets, O., Arokoski, J.P., Saarakkala, S., Korhonen, R.K., 2017a. Simulation of subject-specific progression of knee osteoarthritis and com- parison to experimental follow-up data: data from the osteoarthritis initiative. Sci.
Rep. 7, 9177.https://doi.org/10.1038/s41598-017-09013-7.
Liukkonen, M.K., Mononen, M.E., Tanska, P., Saarakkala, S., Nieminen, M.T., Korhonen, R.K., 2017b. Application of a semi-automatic cartilage segmentation method for biomechanical modeling of the knee joint. Comput. Methods Biomech. Biomed.
Engin. 20, 1–11.https://doi.org/10.1080/10255842.2017.1375477.
Loening, A.M., James, I.E., Levenston, M.E., Badger, A.M., Frank, E.H., Kurz, B., Nuttall, M.E., Hung, H.-H., Blake, S.M., Grodzinsky, A.J., Lark, M.W., 2000. Injurious me- chanical compression of bovine articular cartilage induces chondrocyte apoptosis.
Arch. Biochem. Biophys. 381, 205–212.https://doi.org/10.1006/abbi.2000.1988.
Lohmander, L.S., Englund, P.M., Dahl, L.L., Roos, E.M., 2007. The long-term consequence of anterior cruciate ligament and meniscus injuries. Am. J. Sports Med. 35, 1756–1769.https://doi.org/10.1177/0363546507307396.
Mamisch, T.C., Trattnig, S., Quirbach, S., Marlovits, S., White, L.M., Welsch, G.H., 2010.
Quantitative T2 mapping of knee cartilage: differentiation of healthy control cartilage and cartilage repair tissue in the knee with unloading—initial results. Radiology 254, 818–826.https://doi.org/10.1148/radiol.09090335.
Matzat, S.J., McWalter, E.J., Kogan, F., Chen, W., Gold, G.E., 2015. T2 relaxation time quantitation differs between pulse sequences in articular cartilage. J. Magn. Reson.
Imaging 42, 105–113.https://doi.org/10.1002/jmri.24757.
Momersteeg, T.J.A., Blankevoort, L., Huiskes, R., Kooloos, J.G.M., Kauer, J.M.G., Hendriks, J.C.M., 1995. The effect of variable relative insertion orientation of human knee bone-ligament-bone complexes on the tensile stiffness. J. Biomech. 28, 745–752.https://doi.org/10.1016/0021-9290(94)00121-J.
Mononen, M.E., Tanska, P., Isaksson, H., Korhonen, R.K., 2016. A novel method to si- mulate the progression of collagen degeneration of cartilage in the knee: data from the osteoarthritis initiative. Sci. Rep. 6, 21415.https://doi.org/10.1038/srep21415.
Mononen, M.E., Tanska, P., Isaksson, H., Korhonen, R.K., 2018. New algorithm for si- mulation of proteoglycan loss and collagen degeneration in the knee joint: data from the osteoarthritis initiative. J. Orthop. Res. 36, 1673–1683.https://doi.org/10.1002/
jor.23811.
Mootanah, R., Imhauser, C.W., Reisse, F., Carpanen, D., Walker, R.W., Koff, M.F., Lenhoff, M.W., Rozbruch, S.R., Fragomen, A.T., Dewan, Z., Kirane, Y.M., Cheah, K., Dowell, J.K., Hillstrom, H.J., 2014. Development and validation of a computational model of the knee joint for the evaluation of surgical treatments for osteoarthritis. Comput.
Methods Biomech. Biomed. Engin. 17, 1502–1517.https://doi.org/10.1080/
10255842.2014.899588.
Nieminen, M.T., Nissi, M.J., Hanni, M., Xia, Y., 2017. Physical properties of cartilage by relaxation anisotropy. In: Xia, Y., Momot, K. (Eds.), Biophysics and Biochemistry of Cartilage by NMR and MRI. Royal Society of Chemistry, Cambridge, pp. 145–175.
https://doi.org/10.1039/9781782623663.
Niskanen, J.-P., 2006. Aedes - a graphical tool for analyzing medical images [WWW document]. URL.aedes.uef.fi, Accessed date: 21 September 2018.
Nissi, M.J., Töyräs, J., Laasanen, M.S., Rieppo, J., Saarakkala, S., Lappalainen, R., Jurvelin, J.S., Nieminen, M.T., 2004. Proteoglycan and collagen sensitive MRI eva- luation of normal and degenerated articular cartilage. J. Orthop. Res. 22, 557–564.
https://doi.org/10.1016/j.orthres.2003.09.008.
Nissi, M.J., Salo, E.-N., Tiitu, V., Liimatainen, T., Michaeli, S., Mangia, S., Ellermann, J., Nieminen, M.T., 2016. Multi-parametric MRI characterization of enzymatically de- graded articular cartilage. J. Orthop. Res. 34, 1111–1120.https://doi.org/10.1002/
jor.23127.
Orozco, G.A., Tanska, P., Mononen, M.E., Halonen, K.S., Korhonen, R.K., 2018. The effect of constitutive representations and structural constituents of ligaments on knee joint mechanics. Sci. Rep. 8, 2323.https://doi.org/10.1038/s41598-018-20739-w.
Paproki, A., Engstrom, C., Chandra, S.S., Neubert, A., Fripp, J., Crozier, S., 2014.
Automated segmentation and analysis of normal and osteoarthritic knee menisci from magnetic resonance images–data from the osteoarthritis initiative. Osteoarthr.
Cartil. 22, 1259–1270.https://doi.org/10.1016/j.joca.2014.06.029.
P.O. Bolcos, et al. Clinical Biomechanics xxx (xxxx) xxx–xxx
9
Pizzolato, C., Reggiani, M., Saxby, D.J., Ceseracciu, E., Modenese, L., Lloyd, D.G., 2017.
Biofeedback for gait retraining based on real-time estimation of tibiofemoral joint contact forces. IEEE Trans. Neural Syst. Rehabil. Eng. 25, 1612–1621.https://doi.
org/10.1109/TNSRE.2017.2683488.
Potter, H.G., Jain, S.K., Ma, Y., Black, B.R., Fung, S., Lyman, S., 2012. Cartilage injury after acute, isolated anterior cruciate ligament tear. Am. J. Sports Med. 40, 276–285.
https://doi.org/10.1177/0363546511423380.
Rautiainen, J., Nissi, M.J., Salo, E.-N., Tiitu, V., Finnilä, M.A.J., Aho, O.-M., Saarakkala, S., Lehenkari, P., Ellermann, J., Nieminen, M.T., 2015. Multiparametric MRI assess- ment of human articular cartilage degeneration: correlation with quantitative his- tology and mechanical properties. Magn. Reson. Med. 74, 249–259.https://doi.org/
10.1002/mrm.25401.
Risberg, M.A., Oiestad, B.E., Gunderson, R., Aune, A.K., Engebretsen, L., Culvenor, A., Holm, I., 2016. Changes in knee osteoarthritis, symptoms, and function after anterior cruciate ligament reconstruction. Am. J. Sports Med. 44, 1215–1224.https://doi.
org/10.1177/0363546515626539.
Rodriguez-Vila, B., Sánchez-González, P., Oropesa, I., Gomez, E.J., Pierce, D.M., 2017.
Automated hexahedral meshing of knee cartilage structures–application to data from the osteoarthritis initiative. Comput. Methods Biomech. Biomed. Engin. 20, 1543–1553.https://doi.org/10.1080/10255842.2017.1383984.
Samaan, M.A., Facchetti, L., Pedoia, V., Tanaka, M.S., Link, T.M., Souza, R.B., Ma, C.B., Li, X., 2017. Cyclops lesions are associated with altered gait patterns and medial knee joint cartilage degeneration at 1 year after ACL-reconstruction. J. Orthop. Res. 35, 2275–2281.https://doi.org/10.1002/jor.23530.
Shan, L., Zach, C., Charles, C., Niethammer, M., 2014. Automatic atlas-based three-label cartilage segmentation from MR knee images. Med. Image Anal. 18, 1233–1246.
https://doi.org/10.1016/j.media.2014.05.008.
Stahl, R., Luke, A., Li, X., Carballido-Gamio, J., Ma, C.B., Majumdar, S., Link, T.M., 2009.
T1rho, T2 and focal knee cartilage abnormalities in physically active and sedentary healthy subjects versus early OA patients—a 3.0-tesla MRI study. Eur. Radiol. 19, 132–143.https://doi.org/10.1007/s00330-008-1107-6.
Surowiec, R.K., Lucas, E.P., Fitzcharles, E.K., Petre, B.M., Dornan, G.J., Giphart, J.E., LaPrade, R.F., Ho, C.P., 2014. T2 values of articular cartilage in clinically relevant subregions of the asymptomatic knee. Knee Surgery, Sport. Traumatol. Arthrosc. 22, 1404–1414.https://doi.org/10.1007/s00167-013-2779-2.
Tamez-Pena, J.G., Farber, J., Gonzalez, P.C., Schreyer, E., Schneider, E., Totterman, S., 2012. Unsupervised segmentation and quantification of anatomical knee features:
data from the osteoarthritis initiative. IEEE Trans. Biomed. Eng. 59, 1177–1186.
https://doi.org/10.1109/TBME.2012.2186612.
Van Rossom, S., Smith, C.R., Zevenbergen, L., Thelen, D.G., Vanwanseele, B., Van Assche, D., Jonkers, I., 2017. Knee cartilage thickness, T1ρand T2 relaxation time are related to articular cartilage loading in healthy adults. PLoS One 12, e0170002.https://doi.
org/10.1371/journal.pone.0170002.
Vaziri, A., Nayeb-Hashemi, H., Singh, A., Tafti, B.A., 2008. Influence of meniscectomy and meniscus replacement on the stress distribution in human knee joint. Ann.
Biomed. Eng. 36, 1335–1344.https://doi.org/10.1007/s10439-008-9515-y.
Wang, Y., Wluka, A.E., Jones, G., Ding, C., Cicuttini, F.M., 2012. Use magnetic resonance imaging to assess articular cartilage. Ther. Adv. Musculoskelet. Dis. 4, 77–97.https://
doi.org/10.1177/1759720X11431005.
Wang, A., Pedoia, V., Su, F., Abramson, E., Kretzschmar, M., Nardo, L., Link, T.M., McCulloch, C.E., Jin, C., Ma, C.B., Li, X., 2016. MR T1ρand T2 of meniscus after acute anterior cruciate ligament injuries. Osteoarthr. Cartil. 24, 631–639.https://doi.org/
10.1016/j.joca.2015.11.012.
Wellsandt, E., Gardinier, E.S., Manal, K., Axe, M.J., Buchanan, T.S., Snyder-Mackler, L., 2016. Decreased knee joint loading associated with early knee osteoarthritis after anterior cruciate ligament injury. Am. J. Sports Med. 44, 143–151.https://doi.org/
10.1177/0363546515608475.
Wheaton, A.J., Casey, F.L., Gougoutas, A.J., Dodge, G.R., Borthakur, A., Lonner, J.H., Schumacher, H.R., Reddy, R., 2004. Correlation of T1rho withfixed charge density in cartilage. J. Magn. Reson. Imaging 20, 519–525.https://doi.org/10.1002/jmri.
20148.
Williams, A., Winalski, C.S., Chu, C.R., 2017. Early articular cartilage MRI T2 changes after anterior cruciate ligament reconstruction correlate with later changes in T2 and cartilage thickness. J. Orthop. Res. 35, 699–706.https://doi.org/10.1002/jor.23358.
Wilson, W., van Donkelaar, C.C., van Rietbergen, B., Ito, K., Huiskes, R., 2004. Stresses in the local collagen network of articular cartilage: a poroviscoelasticfibril-reinforced finite element study. J. Biomech. 37, 357–366.https://doi.org/10.1016/S0021- 9290(03)00267-7.
Wilson, W., van Burken, C., van Donkelaar, C., Buma, P., van Rietbergen, B., Huiskes, R., 2006. Causes of mechanically induced collagen damage in articular cartilage. J.
Orthop. Res. 24, 220–228.https://doi.org/10.1002/jor.20027.
Xia, Y., Moody, J.B., Burton-Wurster, N., Lust, G., 2001. Quantitative in situ correlation between microscopic MRI and polarized light microscopy studies of articular carti- lage. Osteoarthr. Cartil. 9, 393–406.https://doi.org/10.1053/joca.2000.0405.
Yang, Z., Fripp, J., Chandra, S.S., Neubert, A., Xia, Y., Strudwick, M., Paproki, A., Engstrom, C., Crozier, S., 2015. Automatic bone segmentation and bone-cartilage interface extraction for the shoulder joint from magnetic resonance images. Phys.
Med. Biol. 60, 1441–1459.https://doi.org/10.1088/0031-9155/60/4/1441.
Yu, H.J., Chang, A., Fukuda, Y., Terada, Y., Nozaki, T., Yoshioka, H., 2016. Comparison of semi-automated and manual segmentation of knee cartilage. Osteoarthr. Cartil. 24, S311.https://doi.org/10.1016/j.joca.2016.01.560.
Zamli, Z., Sharif, M., 2011. Chondrocyte apoptosis: a cause or consequence of osteoar- thritis? Int. J. Rheum. Dis. 14, 159–166.https://doi.org/10.1111/j.1756-185X.2011.
01618.x.
