Skip to main content
NCBI home page
Frontiers in Bioengineering and Biotechnology logoLink to Frontiers in Bioengineering and Biotechnology
. 2026 Apr 9;14:1769107. doi: 10.3389/fbioe.2026.1769107

An inter-laboratory study of the multi-dimensional behaviors of analogue lumbar spine surrogates: towards standardization in spine testing

Emma C Coltoff 1,†, Siril Teja Dukkipati 2,3,†, Philip J Brown 1,*, Mark Driscoll 2,3,*
PMCID: PMC13102567  PMID: 42040771

Abstract

Introduction

Experimental reproduction of lumbar spinal loading throughout activities of daily living is challenging and has historically relied on simplifications to the applied loading to pure moment testing along primary orthogonal anatomical planes. The primary objective of this study was to evaluate the reproducibility and crossplatform translation of multi-dimensional spinal loading protocols between two distinct biomechanical testing systems.

Methods

Lumbar spinal surrogate models are used as repeatable, transportable, and inexpensive specimens to serve as standardized test articles for inter-laboratory comparison. Combined displacement-controlled Flexion-Extension (FE) and Lateral Bending (LB) trajectories and combined Axial Loading and Axial Rotation (AR) trajectories were applied to the specimens on a six-axis robotic arm and a two-axis linear-torsion testing system with custom jig. Both protocols were able to be decomposed and translated between the two testing systems, lending credibility to this method as a standardizable, reproducible process among biomechanics laboratories.

Results

The results demonstrated strong agreement in combined FE-LB range of motion (ROM) and load across the two laboratory setups, with inter-laboratory ROM differences of 0.53°–0.55° (8%–10%) and peak load differences of 0.92–1.64 Nm (10%–18%) for the 3D printed surrogates. Combined AR and axial loading yielded similar behavior, with ROM differences of 0.80°–0.85° (3%–4%) and peak load differences of 0.21–0.89 Nm (1%–5%) between laboratories (all p > 0.05). Furthermore, both laboratories were able to utilize adopted methods to highlight specimen asymmetries and biomechanical responses that would be otherwise overlooked by traditional pure moment testing.

Discussion

This work lays a foundation for future multidimensional testing for capturing complex biomechanical behavior of the spine.

Keywords: 3D printing, analogue, biomechanics, interlaboratory, lumbar, spine, surrogate

1. Introduction

Surrogate models of the lumbar spine have been developed as standardized tools and as substitutes for cadaveric specimens, the current gold standard for biomechanical testing (Domann et al., 2011; Avidano et al., 2008). These surrogates are designed to improve reproducibility across implant testing and between different laboratories, facilitate long-term assessments that are difficult to perform using biological tissue, and mitigate challenges posed by logistics of specimen shipping and ethical approvals (Wheeler et al., 2011). Thus, the repeatability, transportability, and durability of these models make them ideally suited for experimental testing for the purposes of methodology development.

Validation of spinal surrogate models has historically been conducted through pure moment experimental loading, evaluating the relationship between the spine’s kinematic range of motion (ROM) in response to an applied moment along anatomical planes, i.e., Flexion-Extension (FE) in the sagittal plane, Lateral Bending (LB) in the coronal plane, and Axial Rotation (AR) in the transverse plane. However, human spinal motion in daily life often involves simultaneous combinations of FE, LB, and AR loading, making the simplification of their analysis to individual anatomical planes a potentially incomplete approach for characterization of spinal biomechanics (Holsgrove et al., 2015; Fuller et al., 2012; Cholewicki et al., 1996). This simplification may limit clinical translatability, as many pathological conditions and implant failure modes including asymmetric disc degeneration, facet joint overloading, pedicle screw loosening, and adjacent segment disease occur under combined and off-axis loading rather than isolated planar moments. For example, pure moment testing can underestimate facet and intradiscal stresses compared with more physiologically realistic load conditions that combine compression and bending (Rohlmann et al., 2009). Despite this evidence of the spine’s biomechanics as multiplanar in situ, over 70% of experimental biomechanical characterizations of passive spinal behavior and stability in literature are conducted in pure moment testing exclusively (Costi et al., 2021). Some notable deviations from the pure moment paradigm include simultaneous application of a bending moment and an axial compressive force (follower load) hypothesized to replicate the stabilizing effect of spinal muscle activation (Gardner-Morse and Stokes, 2004; Tushak et al., 2023). It is important to note that this constant, curvature-following compressive preload differs from the variable axial loading applied in the present study, in which compressive or tensile forces between the specimen mounting points are actively modulated during motion to interrogate load-dependent mechanical behavior of the spinal construct. Others have employed simultaneous linear ramping or sinusoidal combinations of FE/LB/AR (Schmidt et al., 2007; Schmidt et al., 2008; Spenciner et al., 2006; Zengerle et al., 2021).

While pure moment testing leaves the 90° sector between the FE (sagittal) and LB (coronal) planes untested, these studies have begun to close this gap by testing multiplanar loading combinations along planes that are 30° or 45° from pure moment bending axes with combinations of FE and LB or FE, LB, and AR (Daniels et al., 2012; Spenciner et al., 2006). However, the advent of industrial six-axis robots as well as enhancements to linear-torsion systems with custom jigs enable testing along planes of smaller increments than this, providing more resolution in understanding off-axis spinal behavior (Coltoff et al., 2025a). A multidimensional testing methodology introduced by Coltoff et al. leverages this opportunity by applying combinations of FE and LB loading in infinitesimal increments, thereby mapping spinal behavior between the neutral zone (NZ) and elastic zone (EZ) along previously untested planes (Coltoff et al., 2025a). The logic underlying this methodology can also be extended to a combination of axial loading and AR, as an analysis of coupled rotational and compressive/tensile loads on the lumbar spine.

For this study, two multiplanar loading combinations were chosen as testing modalities to represent physiologically relevant spinal motions: 1) a combination of FE and LB, and 2) a combination of AR and axial loading. Coupled sagittal and coronal plane bending has been widely documented in literature as an inherent feature of lumbar spine kinematics (Dukkipati and Driscoll, 2025a; Dukkipati and Driscoll, 2025c; Liebsch and Wilke, 2025; Cholewicki et al., 1996). Functionally, such coupled FE–LB motions occur during coordinated trunk movements involving rhythmic side-to-side trunk movements (e.g., hula-hooping). Similarly, the combination of AR and axial loading was selected based on evidence that compressive preload substantially influences rotational stiffness, neutral zone behavior, and segmental stability of the spine (Gardner-Morse and Stokes, 2004). In vivo, trunk rotation rarely occurs in the absence of compressive loading generated by body weight and muscle activation. Functionally, this loading scenario approximates motions such as seated trunk rotation while reaching posteriorly (e.g., retrieving an object from the back seat of a vehicle). Although many such combinations can be synthesized, care must be taken to preserve experimental practicality when decomposing these combined motions into the orthogonal loads, particularly when using two-axis linear-torsion systems.

Building on this rationale, the objectives of this work were threefold: 1) to assess performance of two physiologically-inspired combined loading protocols using standardized surrogate models, 2) to develop a standardized procedure for implementing these combined loading protocols, and 3) to compare the outcomes from these combined loading protocols across two distinct testing systems.

2. Methods

2.1. Spinal surrogate models

Two L2-L5 lumbar spine surrogate models were utilized for the work in this study: a novel validated 3D-printed surrogate model (n = 3) (Dukkipati and Driscoll, 2024; Dukkipati and Driscoll, 2025b) and a commercially available biomechanical model (Sawbones®, WA, United States, SKU# 3430-25 (4), n = 1) (Zdero et al., 2023) (Figure 1). This investigation was designed as a preliminary proof-of-concept study to assess the mechanical feasibility and inter-laboratory reproducibility of multiplanar loading protocols in standardized surrogate constructs, rather than to establish population-level statistical inference.

FIGURE 1.

Four spinal column models are shown in two pairs. The left pair, labeled sagittal and coronal, displays transparent anatomical models. The right pair, also labeled sagittal and coronal, shows similar models with synthetic muscle layers, spinal instrumentation, and mounting blocks.

Open in a new tab

L2-L5 lumbar spine surrogates tested: (left) 3D-printed and (right) Sawbones composite surrogate.

The 3D printed surrogate was developed by Dukkipati and Driscoll in the Musculoskeletal Biomechanics Research Lab at McGill University (Dukkipati and Driscoll, 2024; Dukkipati and Driscoll, 2025b). It consists of L2 to L5 solid vertebral bodies and their corresponding intervertebral discs (IVDs)– the fibrocartilaginous cushion between vertebrae – which are derived from the open-source human anatomography database “BodyParts3D” (Mitsuhashi et al., 2009) and the manually designed interspinous and the intertransverse ligaments using the anatomical landmarks on the spinous processes between the adjacent vertebrae. The source dataset corresponds to a healthy 22-year-old male (height: 172.8 cm; mass: 65.0 kg). No geometric scaling was applied during model preparation. The resulting construct demonstrated a lumbar lordosis angle of 44.76°, which aligns closely with reported normative values for asymptomatic young adults (47° ± 10.07°) (Kim et al., 2003). Each anatomical component was printed separately on a Form 3 L printer with default settings at a 100 µm layer height (Formlabs Inc., MA, United States) and assembled into a functional spine unit (FSU). The vertebrae were printed in Formlabs Durable v2 resin and the IVDs and ligaments in Formlabs Flexible 80A V1 photopolymer resin. Post-processing steps followed manufacturer recommendations and involved washing the components in 99% isopropyl alcohol solution to remove excess resin and curing under a 405 nm ultraviolet light source. The parts were then bonded together by applying a small quantity of Flexible 80A v1 resin between the surfaces and spot curing them. After all the vertebral levels were assembled using a jig, the whole model was then cured for 60 min in UV light to ensure proper bonding. Three identical versions of this model were created for this study. The L1-S1 versions of these models were validated previously against literature ex vivo (Stubbs, 2011; Rao, 2012) and in silico (Campbell et al., 2016; Rao, 2012) datasets in pure moment.

The construction of the Sawbones model included proprietary composite materials for each of the vertebrae, IVDs, and interspinous ligaments, with grey resin mounting blocks above L2 and below L5. The geometry of these blocks was integrated into the design of the 3D printed surrogate such that the same mounting adapter could be used for both models. Both the 3D printed model and the Sawbones model were dimensionally similar, with the features of the 3D printed model between the blocks being 1 cm shorter than the Sawbones model in the cranio-caudal direction. The comparisons in this study are limited to construct-level mechanical responses rather than intrinsic material equivalence.

2.2. Testing environments

Two laboratories with different testing equipment were employed in this study to compare the repeatability of the multiplanar loading methodology across testing systems. These setups enabled simultaneous FE and LB bending, as well as combined AR bending and axial loading.

In one setup (Wake Forest University), an industrial six-axis robotic arm, KR 300 R2500 Ultra Robot (KUKA Robotics, Augsburg, Germany), is utilized for biomechanical testing (Figure 2). Each L2-L5 surrogate was mounted to custom fixtures. The fixture at the caudal end of the specimen was mounted to a Delta IP68 SI-330-30 six-axis (ATI Industrial Automation Inc., United States) load cell atop a stand rigidly affixed to the floor. The fixture at the cranial end was mounted to the end effector of the KUKA system. Geometrically similar mounting adapters were used across both laboratories to define the construct’s cranio-caudal axis and ensure consistent boundary conditions. Optotrak (Northern Digital Inc., Canada) optical tracking sensors were affixed to each the L3 and L4 vertebral bodies via custom 3D-printed sensor mounts. One sensor each was attached to the metal fixtures at the cranial and caudal ends of the specimen for L2 and L5. simVITRO (Cleveland Clinic BioRobotics, United States) software was used to interpret the specimen’s ROM behavior from the optical tracking sensors and load behavior from the load cell. Four landmark points were digitized for each of the lumbar vertebra to set up a rigid body coordinate system for the vertebral bodies; the L3-L4 FSU was selected as the central location of the joint coordinate system (JCS) on the specimen for simVITRO’s gravity compensation procedure (simVITRO bioRobotics, 2014). Utilizing the defined rigid body coordinate systems, sensor translations and rotations from the Optotrak system were transformed through simVITRO into anatomical bending measures in anterior, lateral, and superior translations (mm) and FE, LB, and AR bending (degrees) relative to the JCS for the segment and individual FSUs of the spine. The ATI load cell measures were processed through simVITRO to determine the anterior, lateral, and superior forces (N) and FE, LB, and AR bending moments (Nm) on the spine segment in response to loading. Pure moments were achieved through active six-degree-of-freedom control using real-time load cell feedback to minimize off-axis forces and moments during testing.

FIGURE 2.

Two photographs showing a robotic testing setup. The left image depicts a robotic arm with labeled parts: end effector, specimen with optical tracking markers, and a load cell. The right image is a close-up of a spine-like specimen with attached optical sensors labeled L2, L3, L4, and L5, mounted on a testing platform.

Open in a new tab

Testing setup for laboratory with six-axis robotic arm. (left) 3D printed specimen on robot, (right) close-up of the specimen with optical tracking sensors.

In the second setup (McGill University), a two-axis linear-torsion testing machine (Electroplus E10000, Instron, MA, United States) was used along with a custom bending jig (Dukkipati and Driscoll, 2024; Dukkipati and Driscoll, 2025b) to enable simultaneous orthogonal linear and rotary load application (Figure 3). The jig consisted of a rotating arm attached to the mounting adapter of L2 vertebra, and a fixed arm attached to the adapter of the L5 vertebra. The adapters allow for rotation of the specimen along its longitudinal axis, making it possible along the sagittal plane, coronal plane, and any plane in between. The rotational axis of the jig was mechanically aligned with the longitudinal midline of the mounting adapters to maintain consistency with the construct axis defined in the robotic setup. A free-moving XY slide ensured no off-axis load in two of the three translational degrees of freedom, while the third translational direction was passively controlled by the linear-torsion head. This XY slide permitted dynamic translation of the specimen during loading, allowing it to self-equilibrate under applied moments and thereby minimizing shear forces and eccentric loading. The XY slide sits on a bi-axial load cell (Dynacell™ 2527-202, Instron, MA, United States, range ±10 kN and ±100 Nm, resolution 0.1 N and 0.05 Nm) measuring the moment and force imparted onto the specimen during testing. Off-axis loads were monitored via the load cell and remained negligible during testing (<5N), consistent with prior validation of this setup (Dukkipati and Driscoll, 2025b; Dukkipati and Driscoll, 2024).

FIGURE 3.

Laboratory testing apparatus for spinal biomechanics features a mounted spine specimen secured between mechanical arms, labeled to indicate rotating arm, fixed arms, XY slide, linear-torsion head, and load cell for force and movement measurement.

Open in a new tab

Testing setups on two-axis linear-torsion testing system setups with the Sawbones specimen for (left) FE and LB, and (right) AR and axial loading.

2.3. Loading protocols

Previous work has validated each of the spinal surrogates through pure moment loading protocols (DiAngelo et al., 2019; Domann et al., 2011; Dukkipati and Driscoll, 2024; Dukkipati and Driscoll, 2025b; Campbell et al., 2012). These studies were used as a reference for designing the multiplanar loading protocol in this work. Two distinct combinations of loads, first described in (Coltoff et al., 2025b), were applied to the surrogate models: FE and LB and AR and axial loading. Both loading schemes were designed using moment-controlled ROM boundaries determined by testing preceding combined loading. In all loading schemes, the cranial end of the specimen was rotated while the caudal end was fixed.

2.3.1. Moment-controlled boundary

For combined FE and LB, a “boundary” of combined FE and LB loads of 7.5 Nm (Costi et al., 2021) was applied 360° about the craniocaudal axis under a moment-controlled loading scheme to capture the ROM boundary of each specimen corresponding to the 7.5 Nm load (Figure 4). This 7.5 Nm safety limit was chosen to reflect the validity of use of the 3D printed models in previous validation studies (Dukkipati and Driscoll, 2024; Dukkipati and Driscoll, 2025b).

FIGURE 4.

Radar chart titled "ROM Boundary By Angle" compares four datasets: 3D Printed – 1 (yellow), 3D Printed – 2 (red), 3D Printed – 3 (gray), and Sawbones (black). The Sawbones line consistently has higher radius values at all angles, while the three 3D printed samples cluster more closely together near the center, indicating lower range of motion boundaries. Circular axis labels range from 0 to 14, with angles marked in degrees around the chart.

Open in a new tab

FE-LB ROM boundaries defined by six-axis robotic arm.

For combined AR and axial loading, a boundary for AR rotations was obtained by pure moment loading of the specimen via a sinusoidal load up to an amplitude of 7.5 Nm at a rate of approximately 0.3°/second.

2.3.2. Combined flexion-extension and lateral bending

The FE-LB boundary was used to define a trajectory of combined FE and LB rotations, known as “rays”, originating from the origin (no-load position) and ending at the ROM boundary in a straight line. A hybrid control scheme, where forces/translations in load-control and moments/rotations where displacement-control was used for this. The rays trajectory approximates standard pure moment bending wherein the spine moves from the neutral zone (NZ) through the elastic zone (EZ) to the end of the spine’s ROM along the pre-defined boundary, then returns back through the NZ to the end ROM in the direction of the explementary angle. This loading is applied along planes defined by steps of 5° about the craniocaudal axis, where the zero-degree axis is at pure Extension (Figure 5). Specimen ROM and loading in 6 DOF were recorded in response to the rays trajectory, with particular interest in FE and LB moments.

FIGURE 5.

Diagram of a vertebra with two intersecting axes labeled sagittal/flexion-extension (vertical, blue) and coronal/lateral bending (horizontal, red), with arrows at 30, 45, and 60 degrees showing combinations of left lateral bending and extension, labeled accordingly.

Open in a new tab

Polar coordinate system definition relative to craniocaudal axis.

To employ the rays trajectory on the two-axis linear-torsion testing machine, the trajectory must be decomposed along individual loading planes relative to the NZ. Individual planes are denoted by a numeric identifier from 1-12 in Figure 6 by their primary angle, and their explementary angle is found by following the line through the NZ. The ROM boundary of each specimen corresponding to 7.5 Nm of moment in each plane was first recorded using the six-axis robotic arm. These points were used to define the displacement-controlled trajectories applied by the uniaxial testing machine via a linear ramp between the end points with a prescribed loading rate of 1°/sec. Loading along each plane began from a neutral position of the spine where all forces and moments on the segment were minimized. Loading was then applied towards extension, ending at the boundary, then back through the neutral position towards flexion, ending of the boundary, returning to the neutral position.

FIGURE 6.

Two line charts on the left display lateral bending rotation and flexion extension rotation in degrees over time, with oscillating patterns. On the right, a vertebra illustration is encircled by twelve colored lines radiating at thirty-degree intervals, numbered one through twelve, forming a rotation sequence labeled with angles zero, ninety, one hundred eighty, and two hundred seventy degrees, showing the loading sequence order.

Open in a new tab

(left) Combined FE and LB trajectory to simulate rays, (right) rays trajectory decomposed into loading planes, with experimental sequence annotated.

The specimen was manually rotated 15° (error of ±0.5°), verified by a digital angle gauge (Klein Tools, IL, United States) after each loading cycle to align with the appropriate loading planes until loading along all 12 planes was completed. Specimens were not removed from the fixture until loading in all planes was completed.

2.3.3. Combined axial rotation and axial loading

The resulting rotations from the AR boundary capture were used to design a linear ramping signal to capture the specimen’s behaviors approaching and crossing through the neutral position from both AR end points. The control signal sweeps from the neutral position, to the end point on the boundary in right AR, back through the neutral position to the end point in left AR, reversing direction back through the neutral position to the end point in right AR again, and finally returning to the neutral position. For the combined loading, this pattern was applied in tandem with an axial load at 11 loading levels, in increments of 25 N from 50 N of tension (−50 N) to 200 N of compression (Figure 7). Specimen loading and ROM in 6 DOF were recorded in response to the rays trajectory, with particular interest in AR moments and superior translations. A schematic representation of both these protocols is provided in Figure 8.

FIGURE 7.

Two charts on the left show time-based plots of axial rotation (top, oscillating between minus ten and ten degrees) and axial load (bottom, ramping up and down between minus one hundred and two hundred newtons). An arrow leads to a three-dimensional graph on the right, plotting axial load against axial rotation in both coronal and sagittal planes with colored trajectories and labeled left and right movements.

Open in a new tab

AR sweeping pattern with stepwise ramping of axial load.

FIGURE 8.

Flowchart illustrating spine specimen testing: images of mounted spinal specimens lead to range of motion boundary capture at 7.5 newton-meters, followed by displacement-controlled test protocols combining flexion-extension, lateral bending, axial rotation, and axial load, generating moment-rotation curves.

Open in a new tab

Summary of the testing and analysis methodology in this study.

2.4. Post-processing

After testing, 6 DOF ROM and load for the L2-L5 spinal segment were transformed into resultant ROM, load, and stiffness values. The data are presented as 3D representations of these values as well as 2D load-displacement responses along the loading planes (for FE and LB combined loading) through custom MATLAB (Mathworks, California, United States) scripts available in the online repository. ARCMap (Arcus Analytica, Waterloo, Canada), was applied to re-parameterize the continuum ROM and load data prior to performing statistical operations on them through Statistical Parametric Mapping (SPM) via the MATLAB package “spm1d” (Pataky, 2012).

3. Results

All four specimens completed both combined FE and LB, and combined AR and axial loading testing at both laboratories. Specimens underwent testing on the six-axis robotic arm first, followed by the uniaxial system. FE and LB testing was conducted before AR and axial loading testing in both labs. Hotelling t-tests (p-value of 0.05) were performed between the laboratories using ARCMap on the moment-rotation behaviors for the individual loading planes. Data were determined to be normal, and no significant differences were found (all p > 0.05).

3.1. Combined flexion-extension and lateral bending

In the FE and LB combined loading, resultant ROM measures around the boundaries for the 3D printed surrogates ranged from 3.9° to 7.8° across testing systems, producing a nearly circular boundary with only slightly larger ROM along planes closer to the sagittal plane (0°–180°) than the coronal plane (90°–270°). Variation in measurements between testing systems were within 0.54° ± 0.01° for each loading plane, with the uniaxial system always producing higher ROM measurements than on the robotic arm (Figure 9). There was also minimal variation the three 3D printed surrogates within each laboratory (uniaxial = 0.51° ± 0.62°, robot = 0.49° ± 0.61°) across all loading planes. The Sawbones specimen produced ROM measures ∼2.5 times larger (11.4°–19.4°) than those of the 3D printed surrogates. Variability between labs was greater for the Sawbones at 2.77° ± 2.64°; the specimen itself may account for more of this variability than the different testing systems or method of load application.

FIGURE 9.

Four polar coordinate plots compare the resultant range of motion (ROM) of lumbar vertebrae specimens, including three 3D printed models and one Sawbones model. Each plot displays two lines: gold for the robot method and red for the uniaxial method. The diagrams show generally similar but not identical ROM profiles between the two methods, with the Sawbones model exhibiting the highest ROM values. A central inset illustrates anatomical motion directions: extension, flexion, right lateral bending, and left lateral bending.

Open in a new tab

FE and LB ROM across all specimens.

The rays trajectories were designed from a FE and LB combined boundary with a resultant moment of 7.5 Nm. Resultant loads in response to the rays trajectory were therefore expected to be approximately 7.5 Nm. However, for the 3D printed surrogates, resultant loads ranged from 6.74 Nm to 11.52 Nm between the two labs, illustrating some variability in response to loading. Resultant loads for the Sawbones specimen were lower, ranging from 3.5 to 8.4 Nm. Loading was higher across all slices for testing on the robotic arm for all specimens. Across three of the four specimens (excluding 3D printed Surrogate 3), loads were higher for loading planes oriented on the right side of the coronal plane and the extension side of the sagittal plane across both labs (Figure 10), indicating stiffer behavior of the specimen in these regions.

FIGURE 10.

Four line graphs compare resultant moment versus loading plane angle for 3D printed specimens one, two, three, and a sawbones specimen, each showing higher values for robot (gold) than uniaxial (red); a central diagram with spinal illustrations indicates load transitions for flexion, extension, LLB, and RLB.

Open in a new tab

FE and LB peak resultant load across all specimens.

The resultant load and stiffness data across all loading planes were also translated into 3D heatmaps (Supplementary Figures S1–S4), enabling visualization of the change in load and stiffness throughout the ROM in addition to behavior at the boundary. This further illustrates biomechanical differences in planar loading along planes beyond the sagittal and coronal planes.

These data were further able to be divided into heatmaps of outward/loading and inward/unloading behaviors (Supplementary Figure S5). Hysteresis area, or the difference between inward and outward behaviors, was also assessed. This area was relatively consistent across most of the 12 loading planes for the three 3D printed surrogates, with a peak around 15° (mostly Extension with a little Left LB to mostly Flexion with a little Right LB) and 165° (mostly Flexion with a little Right LB to mostly Extension with a little Left LB). Surrogate 3 deviated from Surrogates 1 and 2 between 45° (equal FE and LB) and 120° (slightly more LB than FE). Hysteresis area for the Sawbones was largely different across anatomical loading planes between the two labs.

Off-axis loads were monitored in both testing environments to assess potential parasitic loading effects. In the robotic setup, secondary forces moments remained below 5 N and 1 Nm relative to the 7.5 Nm primary boundary moment across loading planes. In the linear–torsion system, transverse shear forces remained below 5 N throughout testing in all directions, corresponding to an off-axis moment of less than 0.92 Nm. These values were small relative to the applied primary moments and did not demonstrate systematic directional bias. Accordingly, no significant increase in observed primary stiffness was attributed to off-axis loading in either platform, consistent with the absence of significant differences in moment–rotation behavior between laboratories as determined by Hotelling t-tests (p > 0.05).

3.2. Combined axial rotation and axial loading

In the combined AR and axial loading, maximum left and right AR for the testing boundary varied little between the three 3D printed surrogates (10.56° ± 1.28° (left) to 11.84° ± 0.20° (right)). The Sawbones specimen had a smaller AR ROM (5.8° (left) to 6.6° (right)).

AR moment was analyzed across the axial loading force levels (50 N in tension to 200 N in compression). As discussed previously, the relationship between planar AR loading and an axial preload has been studied with respect to AR ROM and moment at the end points of loading or inflection points marking transitions between the NZ and EZ. The data from these combined loading tests can be stitched together to map the changes on AR moment in relationship to changing AR ROM and axial loading (Figure 11). Minimal variation by level in peak load was observed for 3D printed surrogates (8.92 Nm ± 0.77 Nm (left), 9.13 Nm ± 0.97 Nm (right)). These loads were notably higher than the applied 7.5 Nm to obtain the AR ROM boundary, even at the 0 N axial loading state. A slight trend was present in only a few surrogates for either left or right AR wherein AR moment increased with increasing compression, indicating that axial (compressive or tensile) loading may not significantly influence AR ROM in these surrogates.

FIGURE 11.

Heatmap matrix compares torque as a function of axial force and rotation for Robot (left column) and Uniaxial (right column) testing across three 3D printed samples and Sawbones (rows). Color gradient indicates torque from 0 Newton-meters, orange, to 10 Newton-meters, purple, with generally similar patterns across systems, but subtle differences, especially in the Sawbones Uniaxial subplot where the torque distribution is shifted diagonally.

Open in a new tab

AR-Axial Loading heatmaps, by surrogate (rows) and robot (left) vs. uniaxial (right). T = Tension, C = Compression, AR = Axial Rotation.

To further quantify agreement between testing platforms beyond visual comparison of curves, inter-laboratory absolute differences were calculated for ROM and peak load in both loading protocols across all specimens. Differences are reported as mean ± standard deviation and as a percentage of total ROM or peak load for each specimen (Table 1). For the three 3D printed (3DP) surrogates, inter-laboratory ROM differences remained below 10% in both protocols. Peak load differences were similarly small for AR + axial loading (<5% for all surrogates) and below 20% for FE + LB. Larger differences were observed for the Sawbones specimen, consistent with its greater overall variability.

TABLE 1.

Inter-laboratory differences observed in each specimen in both protocols. Absolute differences followed by difference as a percentage of total ROM/Load in parentheses. Differences greater than 10% are highlighted in orange.

ROM (deg.) 3DP-1 3DP-2 3DP-3 Sawbones
FE + LB 0.55 ± 0.07 (8.88%) 0.54 ± 0.08 (8.56%) 0.53 ± 0.11 (9.77%) 2.90 ± 2.44 (20.88%)
AR + Axial load 0.80 ± 0.02 (3.50%) 0.83 ± 0.02 (3.50%) 0.85 ± 0.06 (4.09%) 1.21 ± 0.86 (9.81%)
Peak Load (Nm) 3DP-1 3DP-2 3DP-3 Sawbones
FE + LB 1.62 ± 0.60 (17.38%) 1.64 ± 0.64 (18.36%) 0.92 ± 0.36 (9.98%) 1.55 ± 0.95 (27.23%)
AR + Axial load 0.67 ± 0.07 (3.74%) 0.21 ± 0.08 (1.18%) 0.89 ± 0.13 (4.89%) 3.69 ± 1.53 (24.22%)
Open in a new tab

4. Discussion

This novel study introduces a new paradigm of multiplanar experimental testing for the spine and also presents a methodology for conducting such testing on in both robotic and load frame-based testing environments. Other studies have conducted multiplanar testing with combined FE and LB, as well as AR pure moment loads (Daniels et al., 2012; Spenciner et al., 2006) and have looked at relationships between axial loading and pure moment loading (Sun et al., 2022; Cripton et al., 2000; Bell et al., 2018; Cai et al., 2020; Patwardhan et al., 2003; Zirbel et al., 2013), particularly axial loading as a “preload” prior to experimental testing or a follower load throughout a multi-segment spinal specimen. Yet few existing studies (Holsgrove et al., 2018) evaluate equivalence in performance between different spine testing systems. This type of work is important to the field, as it improves the ability of spinal researchers to compare results across studies and to work towards more standardized spinal testing.

In general, the loading protocol was well-controlled for replicability between the two laboratories, with each lab maintaining a <2% RMS error in their control signals. All testing was conducted at room temperature. Testing order within individual specimens, wherein combined FE and LB loading, followed by AR and axial loading, was maintained. Other applied loading due to preconditioning or troubleshooting was aimed to be minimized within a testing session to control effects of these confounding factors. Determination of the combined FE and LB boundary and AR boundary prior to testing enabled creation of repeatable planar loading trajectories within the specimens’ physiological ROM for application across both testing systems. The trajectories were able to both be run continuously on a six-axis robotic arm as well as being decomposed for application on a loading frame capable of linear and rotary load application. This enabled recreation of 3D heatmap characterizations of spinal behavior in combined loadings, providing a full-field understanding of spinal biomechanics.

The requirement to define specimen-specific ROM boundaries on the robotic platform prior to application on the uniaxial system highlights an important consideration for broader adoption (Figure 8). In the present study, robotic boundary characterization was performed to establish safe physiological operating limits for the custom-manufactured 3D printed surrogates, for which no manufacturer-provided ROM specifications were available, unlike the Sawbones specimen. While the robotic system enables efficient identification of continuous multiplanar boundaries, it is not intended to be a permanent prerequisite for inter-laboratory implementation. Once boundary trajectories are characterized, whether through robotic mapping or manufacturer-provided limits, they may be shared across laboratories as standardized reference maps. A laboratory equipped solely with a linear–torsion platform could then reconstruct a down-sampled boundary by connecting end-range ROM values across rays and iteratively refining the trajectory within physiological limits. Thus, the use of robotic instrumentation in this study reflects a limitation of the 3D printed specimens rather than the proposed protocol itself.

Variability in loads from the applied 7.5 Nm boundary was noted in both combined FE and LB loading as well as AR and axial loading. Moment-controlled loading was utilized to locate these boundaries, while hybrid-control was used for applied the subsequent combined loading trajectories. The alternation between loading methodologies may have contributed to some of this variability, by introduction of coupled loading present due to uncontrolled DOF (Goel et al., 1995).

Despite actions taken for reproducibility, combined FE and LB resultant loads were markedly higher for testing on all samples on the robotic arm and a larger ROM for uniaxial testing system. The specimens were noted to have mechanically changed, especially the Sawbones sample, between the testing laboratories. Delamination damage was observed on the Sawbones sample between the L2-L3 level due to accrual of degradation from prior testing on the specimen, perhaps also contributing to the presence of bilateral asymmetries in LB and AR. Furthermore, though specimens were stored in UV protectant bags and obscured from overhead lights while not being tested and while being transported between the two testing laboratories, exposure to UV lighting and airport X-ray security scanning during transit may have impacted the materials present in these photo-polymeric surrogates. Additionally, friction within the passive XY slide of the linear–torsion system may have introduced minor resistive forces during coupled motion and load reversals, potentially contributing to hysteresis differences. These XY frictional forces could not be quantified with the current bi-axial load cell in the linear-torsion system. Transverse shear forces (Z), however, remained below 5 N (<0.92 Nm off-axis orthogonal moment) during all the tests.

The combined FE and LB loading highlighted additional asymmetries in the 3D printed surrogates, as well. The 3D printed surrogates were derived from real patient anatomy which maintained any patient-specific spinal asymmetries. Unlike pure moment testing alone, loading along the additional loading planes every 15° informs asymmetries that appear in spinal movements that are not exclusively along the sagittal or coronal planes. This helps to capture the reduced left LB in this anatomy, which impacts not only pure left LB but combinations of bending along the sagittal plane in tandem with left LB.

This is also illustrated by the differences in loading along planes between 45° (equal FE and LB) and 120° (slightly more LB than FE), particularly for 3D printed surrogate #3. Surrogate #3 illustrates a similar left LB boundary to surrogates #1 and #2, but a reduced right LB edge to the boundary. Though this resulted in more equivalent resultant load values across all loading planes, it likely illustrates some asymmetries introduced into the surrogate’s behavior due to the fabrication process. Though the primary objective of this work was to compare results of a novel testing methodology between two distinct testing systems, this work may also be used to guide areas of further refinement of these spinal surrogate models.

While numerous studies have established that lumbar FE behavior is influenced by axial follower loads, this study investigated the impact of an axial follower load on AR behavior, for which literature is limited. The results from this study indicated that AR load was not significantly impacted by axial loading. The homogenous composition of the 3D printed surrogate and Sawbones surrogate IVDs may have contributed to this phenomenon, as cadaveric lumbar IVDs have a heterogeneous structure in which the annulus fibrosus within the IVD adapts to increasing axial load to protect the nucleus pulposus (Nie et al., 2024). Future cadaveric lumbar spine testing should be conducted to verify this hypothesis.

The small sample size (n = 4 surrogates) limits the statistical power of this study and restricts the generalizability of the findings. Only a single commercially available Sawbones specimen was tested due to limited availability and subsequent market discontinuation, which precluded intra-specimen comparisons for that analogue. Although multivariate statistical comparisons did not demonstrate significant inter-laboratory differences, the study may be underpowered to detect subtle effects, particularly specimen-specific asymmetries or small stiffness variations. As such, the results should be interpreted as preliminary validation of methodological equivalence. Future investigations should include a larger number of specimens. Nevertheless, the results from this study provide a vast pool of data from which previously undocumented aspects of spinal behavior can be analyzed. An additional structural limitation is that the anterior and posterior longitudinal ligaments (ALL and PLL) were not explicitly modeled as discrete structures in the 3D printed surrogates (Figure 1). Segmental stiffness was instead achieved through the composite material properties of the intervertebral disc region and surrounding printed structures, representing an anatomical simplification. As the ALL is a primary stabilizer in extension, its absence as a distinct structure may influence extension-dominant mechanics. Incorporation of anatomically discrete ligamentous elements remains an area of improvement in these 3D printed models.

5. Conclusion

This first-of-its-kind inter-laboratory study validated two novel combined-loading protocols, characterizing the multiple dimensions of spinal behavior, on two commonly used biomechanical testing systems. Both protocols were able to be decomposed and translated from a six-axis robotic arm to a two-axis linear-torsion testing system, lending credibility to this method as a standardizable, reproducible process among biomechanics laboratories. Spinal surrogate models aided this work by providing a repeatable, transportable, and inexpensive medium for testing between laboratories located over 1,400 km apart in a span of 2 months. The results of this work demonstrate a strong agreement in spinal ROM and resultant load across the two laboratory setups, despite numerous possible sources of variability. Furthermore, both laboratories were able to utilize the methods to highlight specimen asymmetries and biomechanical responses that would be otherwise overlooked by traditional pure moment testing. This work lays a foundation for future multidimensional testing for capturing complex biomechanical behavior of the spine.

Acknowledgements

The authors would like to thank the Orthopaedic Research Society for the ORS Travel Grant.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. This study was funded by the Natural Science and Engineering Research Council of Canada (NSERC, grant no. 250992 and 245375, awarded to MD), Fonds de recherche du Québec (FRQNT, grant no. 315108 and 332139, awarded to SD) and McGill University (MEDA, awarded to SD).

Footnotes

Edited by: Jiaqiu Wang, London South Bank University, United Kingdom

Reviewed by: Yangyang Yang, Shanghai Jiao Tong University, China

James Warren, The University of Manchester, United Kingdom

Data availability statement

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

Author contributions

EC: Validation, Methodology, Data curation, Writing – original draft, Investigation, Conceptualization, Visualization, Software, Funding acquisition, Writing – review and editing, Formal Analysis. SD: Software, Visualization, Data curation, Writing – original draft, Funding acquisition, Formal Analysis, Investigation, Validation, Conceptualization, Methodology, Writing – review and editing. PB: Funding acquisition, Writing – review and editing, Writing – original draft, Project administration, Supervision, Resources, Conceptualization. MD: Writing – review and editing, Project administration, Writing – original draft, Supervision, Resources, Conceptualization, Funding acquisition.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbioe.2026.1769107/full#supplementary-material

DataSheet1.pdf (678.5KB, pdf)

References

  1. Avidano E. E., LaPierre L. J., Friis E. A., James J. E., Johnson A. E. (2008). Synthetic soft tissue characterization of the mechanical analogue lumbar spine. J. Med. Devices 2, 027509. 10.1115/1.2932564 [DOI] [Google Scholar]
  2. Bell K. M., Yan Y., Hartman R. A., Lee J. Y. (2018). Influence of follower load application on moment-rotation parameters and intradiscal pressure in the cervical spine. J. Biomechanics 76 (July), 167–172. 10.1016/j.jbiomech.2018.05.031 [DOI] [PubMed] [Google Scholar]
  3. Cai X.-Y., YuChi C.-X., Du C.-F., Mo Z.-J. (2020). The effect of follower load on the range of motion, facet joint force, and intradiscal pressure of the cervical spine: a finite element study. Med. and Biol. Eng. and Comput. 58 (8), 1695–1705. 10.1007/s11517-020-02189-7 [DOI] [PubMed] [Google Scholar]
  4. Campbell J. R., Imsdahl S. I., Ching R. (2012). Evalution of a synthetic L2-L5 spine model for biomechanical testing. Available online at: https://api.semanticscholar.org/CorpusID:51847461 (Accessed March 15, 2026). [Google Scholar]
  5. Campbell J. Q., Coombs D. J., Rao M., Rullkoetter P. J., Petrella A. J. (2016). Automated finite element meshing of the lumbar spine: verification and validation with 18 specimen-specific models. J. Biomechanics 49 (13), 2669–2676. 10.1016/j.jbiomech.2016.05.025 [DOI] [PubMed] [Google Scholar]
  6. Cholewicki J., Crisco J. J. I. I. I., Oxland T. R., Yamamoto I., Panjabi M. M. (1996). Effects of posture and structure on three-dimensional coupled rotations in the lumbar spine: a biomechanical analysis. Spine 21 (21), 2421–2428. 10.1097/00007632-199611010-00003 [DOI] [PubMed] [Google Scholar]
  7. Coltoff E. C., Hezrony B. S., Marcet P. A., Wilson J. L., Brown P. J. (2025a). A novel methodology for improving understanding of the multiplanar kinematics of the human cervical spine. J. Eng. Sci. Med. Diagnostics Ther. 8, 041107. 10.1115/1.4067767 [DOI] [Google Scholar]
  8. Coltoff E., Dukkipati S., Loss J., Wahbeh J. (2025b). “A multi-laboratory study towards standardizing spine testing: evaluating the impact of protocol and testing environment on spine biomechanics.” Preprint. [Google Scholar]
  9. Costi J. J., Ledet E. H., O’Connell G. D. (2021). Spine biomechanical testing methodologies: the controversy of consensus vs scientific evidence. JOR SPINE 4 (1), e1138. 10.1002/jsp2.1138 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Cripton P. A., Bruehlmann S. B., Orr T. E., Oxland T. R., Nolte L. P. (2000). In vitro axial preload application during spine flexibility testing: towards reduced apparatus-related artefacts. J. Biomechanics 33 (12), 1559–1568. 10.1016/s0021-9290(00)00145-7 [DOI] [PubMed] [Google Scholar]
  11. Daniels A. H., Paller D. J., Feller R. J., Thakur N. A., Biercevicz A. M., Palumbo M. A., et al. (2012). “Examination of Cervical Spine Kinematics in Complex, Multiplanar Motions after Anterior Cervical Discectomy and Fusion and Total Disc Replacement.” full length article. Int. J. Spine Surg. 6 (January), 190–194. 10.1016/j.ijsp.2012.07.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. DiAngelo D. J., Hoyer D., Chung C. (2019). Biomechanical evaluation of a full-length (T12-S) synthetic lumbar spine model. MedCrave Online J. Appl. Bionics Biomechanics 3 (3), 70–75. 10.15406/mojabb.2019.03.00106 [DOI] [Google Scholar]
  13. Domann J., Mar D., Johnson A., James J., Friis E. (2011). The analogue spine model: the first anatomically and mechanically correct synthetic physical model of the lumbar spine. Spine J. 11 (10), S155–S156. 10.1016/j.spinee.2011.08.373 [DOI] [Google Scholar]
  14. Dukkipati S. T., Driscoll M. (2024). Design improvements and validation of a novel fully 3D printed analogue lumbar spine motion segment. J. Bionic Eng. 21 (3), 1388–1396. 10.1007/s42235-024-00512-8 [DOI] [Google Scholar]
  15. Dukkipati S. T., Driscoll M. (2025a). An integrated rigid-flexible body dynamic approach to computationally efficient musculoskeletal modeling and muscle recruitment simulation of the Lumbosacral spine and torso. IEEE Trans. Bio-Medical Eng. 1–14. 10.1109/TBME.2025.3617301 [DOI] [PubMed] [Google Scholar]
  16. Dukkipati S. T., Driscoll M. (2025b). Development and biomechanical evaluation of a 3D printed analogue of the human lumbar spine. 3D Print. Med. 11 (1), 3. 10.1186/s41205-025-00249-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Dukkipati S. T., Driscoll M. (2025c). Evaluation of a fast-solving rigid body spine model inclusive of intra-abdominal pressure. IEEE Trans. Bio-Medical Eng. 72 (10), 3108–3119. 10.1109/TBME.2025.3561692 [DOI] [PubMed] [Google Scholar]
  18. Fuller A. M., Chui J. M., Cook D. J., Yeager M. S., Gladowski D. A., Cheng B. C. (2012). Verification of pure moment testing in a multi–degree of freedom spine testing apparatus. Int. J. Spine Surg. 6 (December), 1–7. 10.1016/j.ijsp.2011.12.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Gardner-Morse M. G., Stokes I. A. F. (2004). Structural behavior of human lumbar spinal motion segments. J. Biomech. 37 (2), 205–212. 10.1016/j.jbiomech.2003.10.003 [DOI] [PubMed] [Google Scholar]
  20. Goel V. K., Wilder D. G., Pope M. H., Thomas Edwards W. (1995). Controversy biomechanical testing of the spine: load-controlled Versus displacement-controlled analysis. Spine 20 (21), 2354–2357. 10.1097/00007632-199511000-00017 [DOI] [PubMed] [Google Scholar]
  21. Holsgrove T. P., Nayak N. R., Welch W. C., Winkelstein B. A. (2015). “Advanced Multi-Axis Spine Testing: clinical Relevance and Research Recommendations.” articles. Int. J. Spine Surg. 9 (January), 34. 10.14444/2034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Holsgrove T. P., Amin D. B., Pascual S. R., Ding B., Welch W. C., Gheduzzi S., et al. (2018). The equivalence of multi-axis spine systems: recommended stiffness limits using a standardized testing protocol. J. Biomech. 70 (March), 59–66. 10.1016/j.jbiomech.2017.09.010 [DOI] [PubMed] [Google Scholar]
  23. Kim W., Kang J., Yeom J. S., Kim K. H., Jung Y. H., Lee S. H., et al. (2003). A comparative analysis of sagittal spinal balance in 100 asymptomatic young and older aged volunteers. J. Korean Soc. Spine Surg. 10 (December), 327. 10.4184/jkss.2003.10.4.327 [DOI] [Google Scholar]
  24. Liebsch C., Wilke H.-J. (2025). Coupled motions of the spine under standardized in vitro conditions: a systematic review and meta-analysis. Front. Bioeng. Biotechnol. 13, 1686524. 10.3389/fbioe.2025.1686524 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Mitsuhashi N., Fujieda K., Tamura T., Kawamoto S., Takagi T., Okubo K. (2009). BodyParts3D: 3D structure database for anatomical concepts. Nucleic Acids Res. 37 (Database issue), D782–D785. 10.1093/nar/gkn613 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Nie M., Huang Z., Cheng R., Zhang Q., Zhang J. P., Fu L. J., et al. (2024). Shape and size of the annulus fibrosus excision alters the biomechanics of the intervertebral disc. Glob. Spine J. 30, 1895–1904. 10.1177/21925682241270090 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Pataky T. C. (2012). One-dimensional statistical parametric mapping in python. Comput. Methods Biomechanics Biomed. Eng. 15 (3), 295–301. 10.1080/10255842.2010.527837 [DOI] [PubMed] [Google Scholar]
  28. Patwardhan A. G., Havey R. M., Carandang G., Simonds J., Voronov L. I., Ghanayem A. J., et al. (2003). Effect of compressive follower preload on the flexion–extension response of the human lumbar spine. J. Orthop. Res. 21 (3), 540–546. 10.1016/S0736-0266(02)00202-4 [DOI] [PubMed] [Google Scholar]
  29. Rao M. (2012). Explicit finite element modeling of the human lumbar spine. Available online at: https://digitalcommons.du.edu/etd/906 (Accessed March 15, 2026). [Google Scholar]
  30. Rohlmann A., Zander T., Rao M., Bergmann G. (2009). Realistic loading conditions for upper body bending. J. Biomechanics 42 (7), 884–890. 10.1016/j.jbiomech.2009.01.017 [DOI] [PubMed] [Google Scholar]
  31. Schmidt H., Kettler A., Heuer F., Ulrich S., Claes L., Wilke H.-J. (2007). Intradiscal pressure, shear strain, and fiber strain in the intervertebral disc under combined loading. Spine 32 (7), 748–755. 10.1097/01.brs.0000259059.90430.c2 [DOI] [PubMed] [Google Scholar]
  32. Schmidt H., Heuer F., Wilke H.-J. (2008). Interaction between finite helical axes and facet joint forces under combined loading. Spine 33 (25), 2741–2748. 10.1097/BRS.0b013e31817c4319 [DOI] [PubMed] [Google Scholar]
  33. simVITRO bioRobotics (2014). Universal musculoskeletal simulator spine module user manual 2012RB-028-048.D. [Google Scholar]
  34. Spenciner D., Greene D., Paiva J., Palumbo M., Crisco J. (2006). The multidirectional bending properties of the human lumbar intervertebral disc. Spine J. 6 (3), 248–257. 10.1016/j.spinee.2005.08.020 [DOI] [PubMed] [Google Scholar]
  35. Stubbs J. (2011). Use of a multi-axis robotic testing platform to investigate the sagittal mechanics of the multi-body lumbar spine. 10.21007/etd.cghs.2011.0302 [DOI] [Google Scholar]
  36. Sun Z., Lu T., Li J., Liu J., Hu Y., Mi C. (2022). A finite element study on the effects of follower load on the continuous biomechanical responses of subaxial cervical spine. Comput. Biol. Med. 145 (June), 105475. 10.1016/J.COMPBIOMED.2022.105475 [DOI] [PubMed] [Google Scholar]
  37. Tushak S. K., Gepner B. D., Forman J. L., Hallman J. J., Pipkorn B., Kerrigan J. R. (2023). Human lumbar spine injury risk in dynamic combined compression and flexion loading. Ann. Biomed. Eng. 51 (6), 1216–1225. 10.1007/s10439-022-03126-5 [DOI] [PubMed] [Google Scholar]
  38. Wheeler D. J., Freeman A. L., Ellingson A. M., Nuckley D. J., Buckley J. M., Scheer J. K., et al. (2011). Inter-laboratory variability in in vitro spinal segment flexibility testing. J. Biomechanics 44 (13), 2383–2387. 10.1016/j.jbiomech.2011.06.034 [DOI] [PubMed] [Google Scholar]
  39. Zdero R., Brzozowski P., Schemitsch E. H. (2023). Biomechanical properties of artificial bones made by sawbones: a review. Med. Eng. and Phys. 118 (August), 104017. 10.1016/j.medengphy.2023.104017 [DOI] [PubMed] [Google Scholar]
  40. Zengerle L., Fleege C., Treuheim T. D., Vogele D., Rauschmann M., Wilke H.-J. (2021). Georg schmorl prize of the German spine society (DWG) 2020: new biomechanical in vitro test method to determine subsidence risk of vertebral body replacements. Eur. Spine J. 30 (5), 1117–1124. 10.1007/s00586-021-06764-w [DOI] [PubMed] [Google Scholar]
  41. Zirbel S. A., Stolworthy D. K., Howell L. L., Bowden A. E. (2013). Intervertebral disc degeneration alters lumbar spine segmental stiffness in all modes of loading under a compressive follower load. Spine J. 13 (9), 1134–1147. 10.1016/j.spinee.2013.02.010 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

DataSheet1.pdf (678.5KB, pdf)

Data Availability Statement

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

Articles from Frontiers in Bioengineering and Biotechnology are provided here courtesy of Frontiers Media SA

RESOURCES