The MRI posterior drawer test to assess posterior cruciate ligament functionality and knee joint laxity

Clinical Magnetic Resonance Imaging (MRI) of joints is limited to mere morphologic evaluation and fails to directly visualize joint or ligament function. In this controlled laboratory study, we show that knee joint functionality may be quantified in situ and as a function of graded posterior cruciate ligament (PCL)-deficiency by combining MRI and standardized loading. 11 human knee joints underwent MRI under standardized posterior loading in the unloaded and loaded (147 N) configurations and in the intact, partially, and completely PCL-injured conditions. For each specimen, configuration, and condition, 3D joint models were implemented to analyse joint kinematics based on 3D Euclidean vectors and their projections on the Cartesian planes. Manual 2D measurements served as reference. With increasing PCL deficiency, vector projections increased significantly in the anteroposterior dimension under loading and manual measurements demonstrated similar patterns of change. Consequently, if combined with advanced image post-processing, stress MRI is a powerful diagnostic adjunct to evaluate ligament functionality and joint laxity in multiple dimensions and may have a role in differentiating PCL injury patterns, therapeutic decision-making, and treatment monitoring.

. Abbreviations: PCL-posterior cruciate ligament. Vector_ASI-vector between the femoral and the tibial axis-surface-intersection. Vector_FT-vector between the apex of the femoral trochlea and the centre of the tibial tuberosity. Vector projections on Cartesian x-axis (i.e., anteroposterior dimension; "x_FT", "x_ASI"), y-axis (i.e., mediolateral dimension; "y_FT", "y_ ASI"), and z-axis (i.e., craniocaudal dimension; "z_FT", "z_ASI") are given, too.  (Table 2), absolute differences versus the unloaded PCL intact condition are detailed as a function of condition and configuration. The corresponding post-hoc test results are given in Supplementary Tables 1 and 2, while the absolute values of each  measure are presented in Supplementary Table 3. Significant differences as a function of PCL condition and configuration were found for the 3D Euclidean vectors vector_FT (that connects the apex of the femoral trochlea [FT] and the centre of the tibial tuberosity [TT]) and vector_ASI (that connects the femoral [fASI] and tibial axis-surface-intersections [tASI]) as well as their projections on the x-, y-, and z-axis as detailed below (see Image Post-Processing to Quantify Joint Laxity). Overall, significant differences were found primarily along the y-axis (anteroposterior dimension: y_ASI and y_FT [both p ≤ 0.001]), and along the x-axis (mediolateral dimension: x_FT [p = 0.008]). In the following, findings of statistical significance refer to the post-hoc comparisons to the PCL intact condition.
For the PCL intact condition, loading induced moderate changes in the joint with the largest and significant increases along the y-axis: Δ 1  In partially PCL-deficient joints, loading-induced changes in the joint were ambiguous and variable for vector lengths and their projections on the x-, y-, and z-axes. Along the y-axis, increases were moderate, yet not significant for y_ASI (Δ 1 [y_ASI] = 9.8 ± 5.8 mm [ns]) or y_FT (Δ 1 [y_FT] = 4.4 ± 5.7 mm [ns]).
In completely PCL-deficient joints, loading induced even larger and partially significant changes in the joint, with the largest increases found along the y-axis (Δ 1 [y_ASI] = 15.2 ± 5.1 mm [***]; Δ 1 [y_FT] = 12.0 ± 5.0 mm [ns]). Mean vector lengths were not homogenously altered under loading. For y_ASI and y_FT, numerous significant differences were noted between the δ 0 -and δ 1 -configurations of the various PCL conditions, while for other 3D computed vector measures, no significant loading-induced differences were found (Table 1 and Supplementary  Table 1).
Overall, the 2D manual reference measurements validated the 3D computed vector measurements ( Table 2). The largest mean increases were found for lPTT (lateral PTT) under loading, e.g., reader 1 [δ 1 ,], PCL intact : 6.6 ± 1.6 mm; PCL partial : 7.7 ± 2.3 mm; PCL complete : 12.7 ± 2.3 mm (p < 0.001). Similarly, AD-MM (i.e., anterior displacement of the medial meniscus), AD-LM (i.e., anterior displacement of the lateral meniscus), PD-MM (i.e., Table 2. Absolute differences of 2D manual reference measures as a function of PCL injury and loading. 2D manual reference measurements as secondary signs of PCL injury were assessed by two readers and as a function of PCL-condition PCL-condition (PCL intact -intact PCL; PCL partial -after partial PCL transection; PCL complete -after complete PCL transection) and loading configuration (  www.nature.com/scientificreports/ posterior displacement of the medial meniscus), and PD-LM (i.e., posterior displacement of the lateral meniscus) changed significantly under loading and as a function of PCL injury (p ≤ 0.001 each).

Discussion
The most important finding of this study was that increasing knee joint laxity may be reliably quantified in situ and as a function of graded PCL injury using the MRI posterior drawer test. Not surprisingly, quantification of altered joint kinematics was particularly relevant in the anteroposterior dimension, even though concurrent loading-induced changes in all dimensions indicate the underlying complexities of functional joint imaging. Clinical-standard MRI techniques evaluate PCL integrity and joint status on a mere morphologic and static level without assessment of ligament function and joint laxity. As morphologic features may mask PCL injuries with pertinent residual laxity and instability, the diagnostic differentiation of partial and complete PCL injuries from the intact PCL is often challenging 9,33,39 . Stress radiography assesses altered femorotibial kinematics in PCL injury, yet suffers from numerous drawbacks such as ionizing radiation, mere 2D projections, and lack of reproducibility and accuracy 16 . As the most powerful contemporary imaging technique for the morphologic evaluation of the knee joint due to its excellent soft tissue contrast, lack of ionizing radiation, and non-invasiveness, MRI is well poised to assess a joint's position and configuration. This study demonstrated that MRI, advanced image post-processing, and standardized loading can be brought together as the MRI posterior drawer test and provides combined morphologic and functional assessment of the PCL and entire knee joint.
Loading was induced by control of pressure using a commercial MRI-compatible loading device previously validated 37 . In line with the manufacturer's instructions, we defined the biomechanical framework conditions in close emulation of the analogous and well-established stress radiographic technique 17,24 . This may facilitate inter-method comparisons and imminent translation of our findings by providing equally efficient and safe framework loading conditions in terms of loading direction and amplitude. The presence of the loading device within the bore required that a body coil be used for imaging instead of a knee coil. While knee coils provide a close fit around the joint and optimized signal-to-noise ratio, the high number of channels of modern body coils compensates for its suboptimal geometry, balances image speed and quality, and may thus fit in well with diagnostic workflows. Yet, systematic comparison of diagnostic quality of 20° of flexion in the knee coil vs. 90° of flexion in the body coil remains to be performed.
Effectual load application became evident by the loading-induced changes in the 3D computed vector and the 2D manual reference measurements.
In PCL-intact joints, vector projections in the anteroposterior dimension, i.e., along the y-axis, increased under loading by 6.6-7.9 mm. Manual reference measurement of lPTT by both readers revealed similar translation of approximately 6.5 mm and overall, these changes indicate the PCL's physiological laxity. Reviewing numerous cadaveric specimen-based biomechanical studies using posterior drawer testing, Kowalczuk et al. reported mean PTTs of 5.4 mm in the PCL-intact joint. Notably, clinical studies suggest lower PTTs in intact joints, e.g. 1.3 ± 1.9 mm 36 . This discrepancy is not surprising as cadaveric specimens are only passively restrained and lack the active tone of musculotendinous structures altogether 19 .
In partially PCL-deficient joints, vector projections in the anteroposterior dimension were variable and ranged from 4.4 to 9.8 mm. Manual measurements of lPTT increased only slightly to a mean of 7.7-7.9 mm which is largely in agreement with the sparse literature 14,18 . Assessing PTT after bundle-wise sectioning and under a posterior force of 134 N, Kennedy et al. reported that transections of either of the PCL's posteromedial and anterolateral bundles resulted in significant, yet only slight PTT increases of 0.9 ± 0.6 mm and 2.6 ± 1.8 mm, respectively 18 . Assessing the contributions of the individual PCL fiber regions that control PTT under loading at 90° of flexion, Covey et al. found that sectioning of the anterior and central fiber regions, while leaving the posterior fiber regions intact, only marginally increased PTT by 0.5 ± 0.4 mm and 0.2 ± 0.0 mm, respectively. Yet, once the posterior fiber regions were transected, PTT was increased substantially by 4.7-6.0 mm 7 . In our study, arthroscopic partial PCL injury was induced by transection of approximately 50% of the PCL fibers from anterior, i.e., through the anteromedial and anterolateral portals. Even though partial PCL injury had to rely on the surgeon's feel, orientational arthroscopic and MRI measurements confirmed the partial PCL to approximately half of the PCL's total diameter. As we did not separate(ly transect) the anterolateral or posteromedial PCL bundles, our setup is comparable to the deficiency of the PCL's anterior and central fiber regions and may explain the only slight and variable increases in PTT. Rather than differentiating the individual contributions of the PCL fibers in biomechanical contexts, we aimed to induce standardized partial PCL injuries as defined by the extent of ligament damage. Yet, despite our best efforts to fully visualize the PCL following synovectomy, standardizing the extent of graded PCL injury proved challenging and transection may have favoured the anterolateral over the posteromedial bundle. The nature of the transection-induced partial PCL injury also limits clinical transferability because actual PCL injury necessarily involves elements of over-stretching and consecutive macro-and microstructural damage of both bundles. This could not be emulated in our experimental setup and would require more refined techniques of joint traumatization. Additionally, subsequent PTT (as determined by partial PCL injury) is not only affected by the number of injured fibers but also by the functionality of the remaining fibers, the bundles affected, the location of ligament injury, and concomitant joint injuries 9,18 . Therefore, partial PCL injuries are inherently variable and, consequently, not uniform in their imaging and clinical manifestations.
Taken together, the small and variable changes in femorotibial kinematics may be the principal reason for the failure of the MRI posterior drawer test to achieve the diagnostic differentiation of partial PCL injury versus other PCL conditions.
In completely PCL-deficient joints, vector projections in the anteroposterior dimension were significantly larger and ranged from 12.0-15.2 mm. Corresponding manual measures of PTT indicated a similar range of 11.8-12.9 mm (for the PTT of the medial and lateral compartment). At 90° of flexion, mean PTT increases of  38 have been reported before, thereby corroborating our findings. While PTT is known to increase substantially in complete PCL injury, its exact amount varies considerably and is affected by study design and methodology, reference measurements, and biomechanical framework conditions. These ill-controlled variables result in the largely variable PTT values of 7.2-18.7 mm reported in earlier studies [7][8][9]38 . The medial and lateral femorotibial compartments behaved differently under loading. In intact and partially PCL-deficient joints, the amount of PTT was substantially higher laterally than medially, i.e., lPTT > mPTT, which may be attributed to the PCL's role in stabilizing the joint against excessive rotation at higher flexion 11,27 . At 90° of flexion, the PCL primarily restrains internal rotation and only secondarily limits external rotation 18 , which explains larger motional translation laterally. In completely PCL-deficient joints, however, PTT was largely similar medially and laterally. Most likely, this finding is due to excessive internal rotation secondary to complete PCL injury which brings about more substantial translation of the medial compartment 18,29 . Consequently, both the mPTT and the lPTT should be determined in manual 2D reference measurements to differentiate partial from complete PCL injury. Beyond each compartment's translation, joint rotation and flexion should be determined, too, to improve measurement validity and diagnostic differentiation of isolated and combined ligament injuries e.g., in efforts to differentiate the isolated PCL injury from the combined PCL and posterolateral corner injuries.
In this study, advanced image post-processing techniques were implemented to parameterize and quantify loading-induced joint changes. Beyond evaluation of the anteroposterior dimension, motional joint changes in the mediolateral and craniocaudal dimensions were assessed, too. Since joints were not constrained under loading, adaptive flexion and rotation was possible which may not be representative of the more confined wholeextremity configuration in patients. Consequently, significant changes in x_FT suggest altered mediolateral alignment and should be considered against this study's experimental design. Nonetheless, such refined techniques provide the basis for enhanced control of otherwise ill-controlled factors such as joint position and rotation that are known to affect PTT quantification 12,18 . Thereby, validity and reproducibility of laxity measurements may be prospectively improved.
Numerous limitations should be recognized that potentially limit clinical applicability and may require further exploration. First, this was an in-situ study with obviously limited translatability to the in-vivo setting as active stabilizers of the joint were absent and only relatively few specimens (n = 11) had been included. Second, the manual 2D or computed 3D measurements were not validated against standard clinical or research measurements, e.g., manual or instrumented laxity measurements or stress radiographic measurements. In our specimens, side-to-side differences were not assessable either. Third, the aged cadaveric specimens might not be representative of the substantially younger clinical population with stronger PCLs. Fourth, our arthroscopic model of PCL injury bears only limited resemblance to the actual PCL injury as detailed above. Similarly, PCL injury often occurs alongside other structural joint, thus manifesting as multi-ligament injury. For example, additional posteromedial and posterolateral corner injuries substantially increase PTT 15 and may challenge correct differentiation of intact, partial or complete PCL injuries 25 . Fifth, our post-processing technique to quantify joint kinematics only provides global estimates of PTT and does not yet allow comprehensive compartmental or regional assessment. In contrast, the 2D manual measures are well-validated imaging measures and have been applied in clinical studies of PCL function that allow assessment of compartmental motional changes 8,9 . For full exploitation of the diagnostic potential of stress MRI, the 3D joint models should be further improved to allow assessment of compartmental translation, joint rotation, and joint flexion. Once additional scientific and clinical data on the loading-induced changes of the joint have been compiled as "(in)stability patterns", these data may be prospectively used to differentiate partial from complete injury as well as isolated from combined injury. Sixth, exact segmentations of the femur and tibia are needed, which, if performed manually, are labour-intensive and time-consuming. For prospective clinical implementation, automated segmentation approaches as suggested previously 35 are necessary. Seventh, the clinical potential of the MRI posterior drawer test is yet unclear and requires further corroboration in clinical studies. The clinically oriented imaging framework (in terms of the clinical 3.0 T MRI scanner, coils, and MRI sequences) and the safe and efficient loading of the joint (by means of the MRI-compatible loading device) provide a solid foundation for the future clinical translation of stress MRI techniques. The next objective is to confirm our findings of physiological laxity in healthy volunteers and to assess aspects of comfort, device handleability and safety, joint fixation, and stabilization as well as measurement validity and reproducibility with the device in clinical operation. Subsequently, the diagnostic potential needs to be assessed in patients (with isolated and combined ligament injuries) and, if possible, in reference to arthroscopy as the reference standard. In these in-vivo studies, the clinical value of the 3D computed vector measures requires additional scientific corroboration.
In conclusion, the MRI posterior drawer test brings together MRI and mechanical loading, and -if complemented by additional advanced image post-processing-it allows for simultaneous assessment of ligament structure and function as well as joint laxity. Consequently, this study provides baseline normative multidimensional imaging markers of knee joint laxity as a function of PCL condition and sets the stage for subsequent in-vivo studies. Beyond the differentiation of acute PCL injury based on femorotibial kinematics, such functional approaches may have a prospective role in therapeutic decision-making and treatment monitoring.

Methods
Study design. This study was designed as an intra-individual comparative in-situ imaging study using knee joint specimens from body donors who had given their written informed consent and had deceased due to conditions unrelated to knee health. Approval by the local institutional review board (Ethical Committee of the Medical Faculty of Heinrich-Heine-University, Düsseldorf, Germany, 2019-682) was obtained before the study. The study has been conducted in accordance with all relevant ethical regulations and local guidelines. were provided by the local Institute of Anatomy (Heinrich-Heine-University Düsseldorf) and included in this study. Details of the donors' history in terms of knee joint injury or surgery was not available. Minimum specimen number was estimated as ten following power analyses on the initial four knee joint specimens 2 . Based on the power of 0.8 and the probability of type-I-error of 0.05, the effect size (defined as the mean paired difference divided by the expected standard deviation) was determined as 1.4 after manually measuring mPTT in the unloaded and loaded configurations of the intact specimens (two-tailed procedure; online software: www. stats todo. com). Assuming larger effect sizes with increasing PCL injury and central or lateral measurements, we decided to include more than the minimum specimen number and, thus, eleven specimens.
Loading device. In line with the manufacturer's instructions, a validated MRI-compatible pressure-controlled loading device was used (Stress Device SE-MR, Telos GmbH, Wölfersheim, Germany) 37 . Mechanistically, the tibia was displaced posteriorly relative to the fixed femur to emulate the MRI posterior drawer test (Fig. 3A1). Fixed with a padded counter-bearing on the distal medial upper thigh (aligned parallel to the tibia), the specimens were placed on the baseplate and in the lateral position at 90° of flexion. A second padded counter-bearing at the distal upper thigh (aligned parallel to the joint line) helped maintain joint flexion and stabilize the knee joint during loading, while the padded pressure applicator was positioned at the level of the tibial tuberosity in loose contact with the joint. In vivo, a third padded counter bearing above the ankle would mechanically fix the tibia, yet, in the present in-situ configuration, the lacking distal lower extremity was compensated for by distal extension of the tibia with a tapered polyvinyl-chloride medullary rod inserted into the medullary cavity and . Following partial synovectomy of the PCL (B 3 ), the PCL was partially transected during the first arthroscopy session by cutting approximately 50% of its cross-sectional area using arthroscopic scissors (B 4 ). The functionality of the remaining PCL fibers was probed (not shown). During the second arthroscopy session, the PCL was completely transected (B 5 ). www.nature.com/scientificreports/ fixed with polymethyl-methacrylate (Technovit-3040, Heraeus-Kulzer, Wehrheim, Germany). Centrally positioned in the scanner's bore (Fig. 3A2), the specimen-loaded device was connected to the control unit located outside of the scanner room (Fig. 3A3). Once the device was pressurized by connecting standard carbon dioxide cartridges (Telos GmbH, liquid mass of 16 g at 50.9 bar, equal to an expanded volume of 8.7 l at 15 °C) to the control unit, forces were directly transferred to the joint by posterior displacement of the padded pressure applicator.

Image data acquisition. Imaging was performed on a clinical 3.0-T MRI scanner (Magnetom Prisma,
Siemens Healthineers, Erlangen, Germany) using an 18-channel body coil (Body 18 SlideConnect, Siemens Healthineers) and a 32-channel spine coil (Spine 32 DirectConnect, Siemens Healthineers) centered on the joint and placed above and below the specimen-loaded device (Fig. 3A2). Each specimen was subject to three MRI measurement series, i.e., intact (PCL intact ), partially transected (PCL partial ), and completely transected (PCL complete ), in two configurations each, i.e., unloaded (δ 0 ) and loaded (δ 1 ). After loading and prior to imaging, 5 min of equilibration were allowed as an empirical time frame that struck a sensible balance between loading-induced tissue adaptation and additional time demand. While the field of view was adapted to the specimen's anatomy, slice geometry and sequence parameters were unchanged between the δ 0 -and δ 1 -configurations. Scout views were used to ascertain constant joint flexion and standardize image conditions for each configuration and MRI measurement series. Acquired for each configuration, the imaging protocol as outlined in Table 3 was performed in line with clinical standard routines and included Proton Density-weighted fat-saturated and T1-weighted sequences. Coronal and axial sequences were used to guide the parasagittal sections along the course of the PCL.
In the intact δ 0 -configuration, a clinical radiologist (SN, eight years of experience in musculoskeletal imaging) confirmed gross integrity of the pertinent intra-and periarticular knee joint structures and evaluated the presence of any signs of pre-existent PCL injury, i.e., aberrant PCL configuration, diameter, or course, focal fiber discontinuity, or excessive PTT 21,41 . These signs had been defined as exclusion criteria, yet were absent in all specimens. Following completion of the imaging protocol at the δ 0 -configuration, the padded pressure applicator was actuated to the pressure level of 2.3 bar, translating to effective posterior forces on the joint of 147 N (= 15 kP) as validated by the manufacturer.
The imaging protocol was obtained for each condition and configuration so that a total of six MRI measurement series were performed per specimen within 48 h. In-between the imaging sessions, the specimens were kept refrigerated at 4 °C and thoroughly warmed to room temperature before scanning. Per specimen, PCL condition, and loading configuration, scanning time was approximately 15 min and, consequently, total magnet time for each specimen was approximately 90 min.
Arthroscopic model of graded PCL injury. Arthroscopic preparation and sequential transection of the PCL was performed by LMW (orthopaedic surgeon with 5 years of experience in arthroscopy) (Fig. 3B1). After establishing access to the joint via the anterolateral and anteromedial portals, the PCL was identified (Fig. 3B2) and synovectomized using curved-tip punch forceps (Arthrex, Naples, FL, US) for optimized visualization prior to transection (Fig. 3B3). To assess pre-existent laxity, the PCL was probed prior to transection. For partial PCL transection, straight-tip arthroscopic scissors (Arthrex) were used to carefully transect approximately 50% of the ligament's diameter at mid substance just cranial to the cruciate ligament intersection, i.e., at mid-substance, and orientational arthroscopic and imaging measurements confirmed the partial PCL injury. Of note, the two individual PCL bundles were not identified or separated during transection (Fig. 3B4). Following partial tran- Table 3. Acquisition parameters of MR sequences. PDW-Proton Density-weighted, fs-fat-saturated, TSEturbospin-echo, SE-spin-echo, w-weighted, cor-coronal, ax-axial, (para)sag-(para)sagittal, n/a-not applicable. (*) aligned to the course of the PCL. Acquisition matrix (pixels) 320 × 320 320 × 320 320 × 320 320 × 320 320 × 320 Reconstruction matrix (pixels) 320 × 320 320 × 320 320 × 320 320 × 320 320 × 320 www.nature.com/scientificreports/ section, the remaining PCL fibers were probed to ascertain functional integrity. For complete PCL transection, the remaining PCL fibers were similarly transected (Fig. 3B5). Following each arthroscopy session, excess fluid was removed and both portals were sutured.
Image post-processing to quantify joint laxity. To assess and quantify joint changes as a function of loading, specimen-specific 3D joint models were implemented using Python software (v3.7.3, Python Software Foundation, Wilmington, Del, US). DK (medical student, 1 year of experience in musculoskeletal imaging) performed the segmentations using the brush tool and polygon mode of ITK-SNAP software (version 3.8.0, Cognitica, Philadelphia, PA, US) 42 . Femoral and tibial bone outlines were manually segmented on sagittal T1-weighted sequences for each specimen, configuration (i.e., δ 0 and δ 1 ), and condition (i.e., PCL intact , PCL partial , and PCL complete ). Based on the outlines of femur and tibia, the coordinates of the following anatomic landmarks were manually registered in Cartesian coordinate systems: (1) most proximal extension of the femoral trochlea (i.e., its tip) that was still covered by articular cartilage (FT), (2) centre of the tibial tuberosity (TT), (3) centers of proximal and distal femur to define the central femoral bone axis, and (4) centers of proximal and distal tibia to define the central tibial bone axis. For (3) and (4), coordinates were selected at the height of the meta-diaphyseal junction and at the most distant diaphyseal extension that was still visualized in the field-of-view. Segmentation outlines and registered coordinates were reviewed for accuracy and, if necessary, corrected, by LMW. To assess intra-reader reproducibility, registration of these coordinates was repeated in a blinded manner on all 11 specimens after 12 weeks. Except for the centre of the TT (2.19 ± 2.02 mm), inter-measurement deviations of coordinates were low and averaged between 0.42 and 0.69 mm. Please see Supplementary Table 4for additional details. Additionally, two fixpoints were automatically computed as the intersections of the central femoral and tibial bone axes and the corresponding articular surfaces of the femoral and tibial cortices, i.e., the femoral and tibial axis-surface-intersections (fASI, tASI).
To quantify loading-induced joint changes in the three dimensions, i.e., anteroposterior (yz-plane), mediolateral (xz-plane), and craniocaudal (xy-plane), 3D Euclidean vectors were calculated between FT and TT, i.e., between (i) and (ii) (vector_FT), and between fASI and tASI (vector_ASI). Additionally, each vectors' magnitude and projections onto the Cartesian x-, y-, and z-axes were determined, too. Based on voxel size (0.6 × 0.6 × 3. Manual reference measurements to quantify joint laxity. For each specimen, configuration, and condition, secondary signs of PCL injury were quantified by means of manual 2D reference measurements obtained by LMW and DK. These signs indicate altered displacement of the tibia versus the femur 9,26 and were measured on the central medial and lateral slices of the respective femorotibial compartments 8,9 using the manual image analysis toolbox of the in-house picture archiving and communications system (Sectra Worksta-tion101, IDS7, Linköping, Sweden). The slices on which the measurements were performed were the same for each set of measurements and the measurements were defined as follows: ( Data are presented as means ± standard deviations. For each condition and configuration, absolute differences were referenced to δ 0 [PCL intact ]. Consequently, absolute differences (Δ x ) were calculated as Δ x [PCL condition ] = δ x [PCL condition ] − δ 0 [PCL intact ]. Assuming normal distributions of the 3D computed vector and 2D manual reference measures, quantitative measures for each configuration and condition were assessed specimen-wise, using repeated measures ANOVA (two-sided) with Bonferroni's test for pairwise post-hoc comparisons. The level of significance was set to p ≤ 0.01 to decrease the number of statistically significant, yet clinically irrelevant findings, and further stratified as 0.01 ≥ p > 0.001 (**) and p ≤ 0.001 (***).

Data availability
The main data supporting the results in this study are available within the paper and its Supplementary Information. Any additional datasets generated and analyzed in this study are available from the corresponding author on reasonable request.