Multi-color and Multi-Material 3D Printing of

Knee Joint models

Oliver Grimaldo Ruiz 1, Yasin Dhaher 2

1. Department of Structural, Geotechnical and Building Engineering (DISEG),

Politecnico di Torino, 10129 Turin, Italy.
Contact: +39 3337933976.
E-mail: [email protected]
URL: http://www.diseg.polito.it/en/personale/scheda/(nominativo)/oliver.grimaldo

2. Physical Medicine and Rehabilitation, Orthopaedic Surgery, UT Southwestern

Medical Center, 5323 Harry Hines Blvd, Dallas, TX 75390, USA.
Contact +1 214-648-0007.
E-mail: [email protected]
URL: https://profiles.utsouthwestern.edu/profile/175158/yasin-dhaher.html

Abstract
Background Every year, 3DP provides more alternatives and solutions in the
medical field. Applications such as custom-made prosthetics and implants,
platforms for pharmaceutical research, and PSAMs are the immediate emerging
trends. Certainly, 3DP advancement is the convergence of multiple factors
including improvements in medical software, 3D printer evolution, availability of
new printing materials, improved industry support, and increasing commitment
from medical societies and regulators. The overarching theme of this study is
centered on exploring possible PSAMs and 3DP applications for improving surgical
outcomes in orthopedics, particularly in ACL-R as well as providing functional
models for TKA.
Methods 3-Matic, Rhinoceros, and SolidWorks were used to create three 3D
computer-generated PSAMs: (1) Knee Joint model, wherein collagen fibers matrix
structure is mimicked, (2) ACL-R model using a BPTB graft, incorporating key
surgical outcomes such as orientations-architecture and positions-dimensions of the
tunnels, as well as a custom-made SG based on patella anatomy (3) TKA model
considering custom-made CS implants with symmetric tibial bearing design.
Before printing, mechanical uni-axial tensile tests of materials were conducted
using an Instron S3300, following the ASTM designation D412-C. The printing
materials selection process and matching with anatomical structures were based on

1
the analysis of the mechanical pattern of the strain-stress curves from different
combinations of Agilus30™. The Stratasys J750™ printer was used to manufacture
the ACL-R model (previous study), the ACL-R model with SG, and the TKA
model.
Results The combinations No. 1-4 were chosen for 3DP with elastic modules of
1.8-0.7 MPa and Pearson coefficients of 0.980-0.991 respectively. The PSAMs
were tested manually simulating 50 flexo-extension cycles without presenting
ruptures, custom-made SG matches perfectly with PT anatomy.
Conclusion Functional PSAMs were printed with high fidelity, considerable cost,
and short duration from planning to manufacturing. These coincided completely
with 3D computer-generated PSAMs replicating fibers and features of the Knee
Joint anatomy. The proposed PSAMs can be considered as an alternative to
replacing cadaver specimens for medical training, pre-operative planning,
education purposes, and validation of predictive models. We highlight the potential
of PolyJet manufacturing combined with specialized medical software as a path to
change the way specialists and researchers plan, execute, and validate complex
procedures.
Keywords Three-dimensional printing, Knee Joint, Patient-specific anatomical
models, Anterior cruciate ligament reconstruction, Total knee arthroplasty.

Background
Additive manufacturing (AM) also widely known as three-dimensional printing
(3DP) is an emerging and revolutionary technology that is getting substantial
interest in several key areas such as the automotive, aerospace, military, and
medical fields. The 3DP process is based on the principle of layered manufacturing,
in which materials are overlapped layer-by-layer enabling the build of 3D objects
(1). Nowadays, the impact of 3DP in the medical field has acquired considerable
relevance in the scientific and academic communities owing to its potential and
wide range of applications. However, the 3DP role in medicine is not recent, this
has been reported since the early 1990s and in recent years, there has been a
considerable rise in the number of emerging trends in the field, demonstrated by the
growing body of literature featuring clinical work and medical research (2). The
3DP advancement is the result of the convergence of multiple factors, including
improvements in medical software, 3D printer evolution, the availability of new

2
printing materials and improved industry support, and increasing commitment from
medical societies and regulators (3). Current research applications are classified
into the following five main areas of focus: (1) Patient-specific anatomical models
(PSAMs), (2) Custom-made prosthetics and implants, (3) Local bioactive and
biodegradable scaffolds, (4) Pharmaceutical research, and (5) Research on directly
printing tissues and organs with complete life functions. Although, such
applications remain far from widespread in clinical use due to several technical and
scientific issues that are currently under study (4).
In particular, PSAMs manufacturing is becoming increasingly popular and
accessible due to its application in pre-operative planning, surgical treatment
analysis, medical training, and education and research purposes (5). The 3D
computer-generated PSAMs are based on digital imaging and communications in
medicine (DICOM) file formats, data derived from several acquisition modalities
such as computed tomography (CT), magnetic resonance imaging (MRI), and
ultrasound (US) (6). The DICOM file formats must be converted into a format,
which can be recognized by the 3D printer. Therefore, first, it is uploaded into a
program (e.g. Mimics, 3D Slicer, OsiriX) which enables segmentation and 3D
reconstruction of the images. It is then exported in a standard triangle language
(STL) file format making it readable by computer-aided design (CAD) and
computer-aided engineering (CAE) software, which are used for design and
simulation. Finally, 3D objects are created by repetitively moving the print head in
the Z-direction and depositing the desired material into layers sequentially (7). The
PSAMs have been utilized extensively in the area of orthopedics, providing
essential information such as sizes, directions, positions, angulations, and
conditions of the bones and surrounding tissues. Researchers and surgeons use this
preliminary knowledge for studying complex cases, teaching students and patients,
rehearsing the procedures in risk-free settings, pre-procedure designing of grafts
and implants, performing finite element analyses, development of simulators, and
examination platforms (8,9).
Multiple authors describe the significant role of the PSAMs in three main domains
of orthopedics: trauma, degenerative disorders, and oncology (10–12). In the Knee
Joint, the most investigations conducted are focused predominantly on total knee
arthroplasty (TKA) and their importance in the development of patient-specific
prostheses (PSP), instrumentation (PSI) and custom-made implants (13–15).

3
Nevertheless, anterior cruciate ligament (ACL) tear remains the most frequently
intra-articular performed surgery in orthopedic trauma (16,17). Despite its
prevalence and impact, the number of publications about the PSAMs and 3DP
applications in ACL reconstruction (ACL-R) has been relatively unexplored. The
design of patient-specific ACL femoral tunnel guide (18) and a method of accurate
bone tunnel placement for ACL-R (19) are the publications highest highlighted.
This statement does not consider other important studies and contributions about
surgical and computational simulations based on 3D computer-generated PSAMs.
The ACL-R aims to restore physiological joint biomechanics in symptomatic knee
instability as well as prevent secondary damage to other structures, such as the
cartilage and menisci (20–22). Critical factors such as post-operative traumatic
injuries, lack of graft incorporation, loss of motion, and surgical technical errors
(surgical technique and surgeon skills) are some of the problems associated with
unsatisfactory medium-long-term clinical outcomes (23,24). Moreover, several
studies have reported the rise of re-rupture cases due to graft failure and the
development of premature degenerative diseases such as osteoarthritis (OA) (25–
30). The ACL-R has undergone major advances in the last years. Nowadays,
different tendon autogenous options exist for the ACL-R. However, the bone-
patellar tendon-bone (BPTB) and the semitendinosus-gracilis (STG) are the most
commonly used and are most successful as alternatives. Each type of graft is
associated with inherent risks and benefits. Therefore, there is no "ideal" graft to
use for ACL-R (31–33). The success or failure of the graft depends on several
surgical parameters including graft stiffness, dimensions and pre-tensioning, tunnel
placement and orientation, and donor-site morbidity (non-modifiable) (34).
Currently, surgical simulations incorporate detailed biomechanical parameters
associated with realistic predictions with the potential to improve clinical outcomes
of the surgery. However, surgical expertise and pre-operative planning are crucial
factors in the ACL-R outcome.
Despite all the recent advances in the technical aspects, traditional ACL-R methods
have proven to be the better option to restore pre-injury activity levels.
Additionally, there is mixed evidence about the effect of ACL-R on the
development of premature OA (35). Structural changes combined with medium and
long-term changes in the dynamic load of patients who have undergone a deficient
ACL-R contribute significantly (36). Thus far, there are no available interventions-

4
treatments to restore degraded structures or decelerate disease development. During
OA progression, the entire joint organ is affected, including cartilage, bone,
synovial tissues, ligaments, and meniscus (37). In an advanced stage, TKA is
considered as the most suggested surgical procedure to restore mechanical axes,
correct alignment, and soft-tissue balance. Although in several cases, TKA has
satisfactory functional and cost-effective outcomes there is significant variability
and globally higher revision rates. Most failures can be attributed to the choice of
replacement components, fixation of the cement-implant interface, incorrect
ligament balance or incorrect alignment, surgical technique and experience, and
post-operative care (38). Due to their potential, the PSAMs and 3DP applications
may improve intra-operative outcomes, problems associated with graft-failure,
recognition of nonanatomic tunnels, surgical technical errors, and reduce the risk
of developing OA of the reconstructed knee.
The overarching theme of this study is centered on exploring possible PSAMs and
3DP applications for improving surgical outcomes in orthopedics, particularly in
ACL-R as well as providing functional models for TKA.
We manufactured and evaluated three multi-color and multi-material Knee Joint
functional models, specifically focusing on the design of a collagen fibers matrix
that mimics the hierarchical structure of specialized connective soft tissues. We
aimed to take advantage of the capability of the Stratasys J750™ printer combined
with medical software Materialise 3-Matic. Our models integrated key surgical
outcomes of the ACL-R computational framework using a BPTB auto-graft and a
custom-made surgical guide for avoiding graft tunnel length mismatch.
Furthermore, we created a model affected by the advanced stage of OA with soft
tissues involvement. In our design, we considered a custom-made cruciate
sacrificing (CS) implants with symmetric tibial bearing design, we adapted and
assembled the PSP components simulating a TKA procedure.
The methodology reported for the production of the multi-material PSAMs might
be considered as a new avenue to develop models for replacing cadaver specimens
for medical training, research and education purposes, pre-operative planning, and
validation of computational predictive models.

Materials and Methods

Multi-color and multi-material three-dimensional printing.

5
The present study was developed in the Shirley Ryan Abilitylab research hospital
which has a Stratasys J750™ (Stratasys, Eden Prairie, MN) multi-color and multi-
material 3D printer. The system is by far one of the most advanced technologies of
3D multi-material printing. This uses PolyJet AM technology to manufacture
highly realistic and functional prototypes with sharp precision, smooth surfaces,
and fine details in a wide range of materials with variable durometers and the
possibility of choosing between 360,000 color combinations.
There are many types of 3D printers available, which use a variety of media,
substrates, and printing technologies. In particular, the Stratasys J750™ uses
ultraviolet (UV) radiation for curing layers of jetted photopolymer. The jetting head
is formed by a matrix of jetting orifices disposed along the Y-axis and is mounted
on a carriage that allows for X forth-and-back displacements and alternate
transverse Y relocations. Subsequently, the manufacturing tray moves in the
vertical Z direction, after each layer has been successfully manufactured (39).
The Stratasys J750™ has a large manufacturing tray; the maximum build size of a
prototype is 490 x 390 x 200 mm (19.3 x 15.35 x 7.9 in.) The system enables
simultaneously mixing up six different materials. In addition to the potential of
adjusting material hardness according to shore A scale, other capabilities including
accuracy of up to 0.2 mm and smaller layer thickness (LT) of 0.014 mm. The system
has three print modes in line with the desired surface finish, production time, and
the number of materials incorporated. (1) High Quality six different materials
(0.014mm LT), (2) High Mix six different materials (0.027mm LT) and (3) High
Speed three different materials (0.027mm LT).
The print materials available have special features such as translucency, flexibility,
resistance to UV rays, high temperature, and deflection. Resins are capable of
simulating properties ranging from rubber-like to transparent, even high toughness.
The system has two main families: Digital model and Model materials, the first one
including engineering plastic acrylonitrile butadiene styrene (ABS) in their versions
Digital ABS Plus - Digital ABS2 Plus™ (main material used in Fused Deposition
Modeling (FDM) technology). The second one includes primary materials options:
Vero™ family (rigid opaque materials), RGD525™ (high-temperature resistant
materials), DurusWhite™ (simulated polypropylene materials), Tango™-
Agilus30™ (rubber-like materials) and VeroClear™ - RGD720 (transparent
materials).

6
The Stratasys J750™ provides many AM advantages, such as the incorporation of
multiple colors and materials in a single project, optimization of printing time, easy
support material removal (waterjet removal), and easy operation.
Image data management.
Three 3D computer-generated PSAMs were created: (1) Knee Joint model (with a
collagen fibers matrix structure), (2) ACL-R model, and (3) TKA model. They were
based on standard triangle language (STL) files corresponding to a PSAM of the
right male Knee Joint (34). The PSAM incorporates patello-femoral (PF) and tibio-
femoral (TF) joints. The following anatomical structures are included: femur, tibia,
patella, fibula, major ligaments, articular cartilage, menisci, retinacula, and patella
and quadriceps tendons (PT-QT). The 3D computer-generated PSAMs were
developed in Materialise 3-Matic (Materialise NV, BE), a design and meshing
software for anatomical data. They were exported in STL format to the CAD
software SolidWorks, (Dassault Systèmes, France) where they were converted to
SolidWorks part file (SLDPRT) format, before being assembled through the same
application. The SolidWorks final assembly format (SLDASM) was compatible
with the Stratasys GrabCAD print™ software of the Stratasys J750™ multi-
material printer where print mode, orientation, and materials were set. The
workflow illustrated in Figure 1. shows the different file formats used in this study.
The printing materials selection has been integrated according to the results of the
mechanical characterization. The format extensions and file names are associated
with a software application, which opens, manages, and saves these types of files
used in this study. These are shown in Table 1.
Table 1: 3D File formats used in the current study. Extensions, file names, and
software applications.

Format extension Filename Software application

7
(.3dm) 3D Object Rhinoceros 3D
(.mxp) Project Materialise 3-Matic
(.SLDPRT) 3D Object SolidWorks
(.SLDASM) Assembly SolidWorks
(.print) Print project GrabCAD Print
(.STL) Standard triangle language Global CAD software

8
Figure 1: Schematic illustration of the formats used in this study. From the Knee Joint model, three approaches are followed: (1) Design of a collagen
fibers matrix that mimics the hierarchical structure of specialized connective soft tissues, (2) development of an ACL-R model integrating key surgical
outcomes of the ACL-R computational framework and a custom-made surgical guide (SG) for avoiding graft tunnel length mismatch, and (3)
development of a model that represents total joint arthroplasty with the integration of PSP components. The illustration shows a path corresponding to
the 3D printing trial with the ACL-R model* used in the previous study.

9
Knee Joint model manufacturing with a collagen fibers matrix structure.
Knee Joint specialized connective soft tissues play a crucial role, providing
strength, transmitting mechanical loads, and contributing to passive support and
stability. Indeed, all these functions are made possible by their hierarchical
organization. In particular, tendons and ligaments share many similar features. They
are load-bearing structures, their high tensile strength ∼100-140 MPa and their
stiffness ~1.0-1.5 GPa provide all the functional requirements associated with
locomotor movement. As expected, both easily bend and change shape to
accommodate changes in joint position and skeletal orientation (40).
Highly paralleled collagen fibrous units characterize tendons and ligaments.
Accordingly, it can be argued that these tissues are analogous to engineering fiber
composites where fibers are laid down in parallel for directional reinforcement (41).
The matrix of collagen fibrils aligned (approximate diameter Ø collagen fibril 1.5
nm) is organized into long cross-striated fibrils that are arranged in bundles to form
fibers (approximate diameter Ø fiber 50-500 nm). Fibers are further grouped in
arrays called fascicles (Ø fascicle 50-300 µm), these arrays together form the
ligament (Ø ligament fiber 0.1-0.5 mm) (42). In the Knee Joint, the hierarchical
structure of connective tissues described determines the mechanical behavior.
Therefore, the knowledge of its mechanical properties is essential to elucidate
behavior and function, as well as for selecting appropriate materials used in surgical
reconstructive procedures.
To mimic the collagen fibers matrix structure, the STL files of the initial Knee Joint
model were exported to Materialise 3-Matic software. A frequent problem in the
3D objects management is the relative position. In general, there is no match
between the global reference system (GRS) of the different applications. The
Materialise 3-Matic software integrates orientation tools (translate & rotate) for
precise positioning of anatomical components according to anatomical references.
The first step in fiber design was to establish the orientation of the Knee Joint about
the anatomical and GRS planes of the application.
A diameter of 0.6 mm was selected, based on approximate diameter for the fiber
(43). A tolerance of 0.1 mm was provided considering a possible expansion of the
material during the printing process. Successively we used a systematic method to
generate contours and paths for each fiber distinct from each other. Fibers were
created along with each structure from traced paths using commands Soft curve &

10
Sweet-loft. Final matrix fibers structure involved virtual post-processing using
Auto-fix, uniform Remesh, Reduce, Smooth & Wrap commands to clean up and
correct surface geometry errors, optimize surface mesh and generate a better-
refined surface finish for final 3DP.

Figure 2: (A) Illustration tracing fibers through the medial collateral ligament
(MCL), contour and sketches are shown (B) collagen fibers set manually created to
MCL (C) Posterior view of the Knee Joint with all created fibers (D) Anterior view
of the Knee Joint, cross-section MCL, the fiber diameter is reported.
The number of designed fibers for cruciate ligaments, collateral ligaments, PT-QT,
medial, and lateral patella-femoral ligaments (MPL-LPL), and medial and lateral
patellar retinacula (MPR-LPR) were reported in Table 2.
Table 2: Number of fibers designed for the Knee Joint Model.

Soft tissues Number of fibers

Anterior cruciate ligament (ACL) 50

Posterior cruciate ligament (PCL) 45
Medial collateral ligament (MCL) 60
Lateral collateral ligament (LCL) 50
Quadriceps tendon (QT) 100
Patella tendon (PT) 45
Medial patella-femoral ligament (MPL) 30
Lateral patella-femoral ligament (LPL) 30
Medial patellar retinacula (MPR) 25
Lateral patellar retinacula (LPR) 20

11
ACL-R model manufacturing and surgical guide for improving surgery
outcomes.
The statistics of post-operative clinical outcomes represent the definitive proof of
success in the treatment of ligament injuries. It is a clinical basis aimed at improving
the results in procedures such as ACL-R. As mentioned, the success of the ACL-R
depends on several surgical factors. Various specialists give strong importance to
the surgical technique, which is associated with an adequate medical training and
accurate pre-operative planning as well as the graft harvest. We sought the
manufacture of a multi-material ACL-R model integrating all key surgical
outcomes of the ACL-R computational framework using a BPTB auto-graft and a
transtibial technique (TT) with single-bundle from a predictive model reported by
(34). The approach incorporated orientations-architecture, position-dimensions of
the femoral and tibial tunnels as well as the design of a custom-made SG based on
the PT anatomy. The SG aimed to solve the problem associated with auto-graft and
tunnel length mismatch. Pre-operative measurements of the BPTB auto-graft length
were performed, specifically, the distance, measured from the origin in the lower
portion of the patella until its insertion in the tibial tubercle. According to (43), if
graft length is greater than or equal to 40 mm, the PT graft is a suitable candidate
for replacement. Several authors suggest an average length of 40 mm for graft and
20 mm for each bone plug. The BPTB block must have a rectangular geometry, a
width of the graft and the bone plugs (tibial and patellar) can range between 9 and
11 mm, a width of 9 mm was chosen. The ACL-R model was developed in the
Materialise 3-Matic software. The measurements and landmarks were made from
the Knee Joint Model with Measure & Landmark commands. Besides, we consider
for the SG design an oscillating saw blade of 0.8 mm wide instead of a traditional
scalpel for the graft harvest. Cut-plans, Boolean tools, marking & extrude
commands were involved in the SG design. Lastly, the graft harvest was simulated
with Boolean tools & Trim commands following the SG dimensions. Femoral and
tibial tunnels of the knee were drilled following the orientations, and dimensions
reported by (34) with 10 mm drills.

12
Figure 3: Schematic illustration of the pre-operative measurements and SG
positioning. (A) Preliminary measures (mm) and landmarks of BPTB autograft, (B)
Simulation of the cut-planes in the graft harvest, (C) Positioned SG, and graft
harvest according to the pre-operative guidelines.

Figure 4: Schematic illustration of the femoral and tibial tunnels architecture,

orientations, dimensions, and positions.

TKA model manufacturing and adjustment of PSP components in the Knee

Joint model.
TKA involves three critical components: femoral, tibial, and articular components.
The femoral component is perhaps the most complex of them. The component has
a convex shape, which emulates the curvatures of the femoral condyles (located at

13
distal femur). The most common cause for premature failure of TKA is aseptic
loosening of articular components. When that occurs, all components fail. The
wrong relation between implant surfaces is usually the main cause of aseptic
loosening. This causes an uneven stress distribution, which leads the component to
the failure.
The multi-material TKA model was developed in the Materialise 3-Matic software,
the Knee Joint model affected by advanced OA stage with soft tissue involvement
was simulated. We chose a custom-made CS implant with symmetric tibial bearing
fixed design. The model represents the adaptation and suitable relationship of the
prosthetic elements following the requirements: (1) the prosthetic components must
have the ability to replicate joint motion as closely as possible. (2) The size of the
implants must be custom-made to the actual anatomy of the Knee Joint.
First, the anatomical and mechanical axes of the bone components were defined.
Pre-operative measurements of the clinical angles were performed using Measure
& Landmark commands: anatomic-mechanic femoral angle (FMAa) anatomic
lateral distal femoral angle (FDLAa) and mechanical lateral distal femoral angle
(FDLMa) (44). Likewise, measurements of the bone components were performed
for the custom-made design of the prosthetic components. Cutting Planes & Trim
commands were used to remove bone and cartilage components as follows: 11 mm
of the proximal tibia (cross-section), 8 mm of the distal femur (cross-section), 7 mm
in the posterior region of the femur (coronal section), 11 mm in the postero-inferior
region of the femur (cross-section) and 6 mm in the anterior region of the femur
(coronal section), ensuring a space of 20 mm for the replacement of the joint
component. The patella was resected 14 mm from the anterior region (coronal
section).

14
Figure 5: Schematic illustration of the pre-operative measurements and simulated
cutting planes. (A) Mechanical and anatomical angles FMAa = 6.09º [5-7º] -
FDLAa = 79.62º [79-83º] -FDLMa = 86.72º [85-90º] with their normal angular
ranges were reported, (B) Anatomical measurements and simulation of the cutting
planes in the bone components, (C) subtraction of the prosthetic components of the
bone component according to the planes.

The femoral and tibial components were exported from Rhinoceros 3D (Robert
McNeel & Associates) where they were designed based on bone measurements
through standard, Planes & Set View commands (Figure 6). The implants were
adapted to the bone components, the tibial and patellar bearings were designed from
them using Boolean & Marking commands. These were adapted to the trajectory
and geometry of the femoral and tibial components respectively. In the neutral
position, the femoral component was aligned, so that the resection of the distal bone
was perpendicular to the mechanical axis of the femur, and the anterior and
posterior resections are parallel to each other (Figure 7).

15
Figure 6: Schematic illustration of the prosthetic components adjustment. Design
components were based on the anatomy of the bone components. Patellar and tibial
bearings were designed based on the femoral component.

Figure 7: Schematic illustration of the TKA model alignment in a neutral position.

(A) Posterior view TKA model, femoral and tibial components are aligned
perpendicular to the mechanical axis. (B) Lateral TKA model, the alignment
corresponded to 90° concerning the mechanical axis. (C) Transverse view TKA
model, the femoral component, anatomic axes-alignment surgical epicondylar axis,
and mechanical axis corresponded to 90°. (D) Isometric view assembly of all the
components of the TKA model.
Mechanical test and the Material selection and matching.
To print the proposed PSAMs, a key aspect was to determine appropriate material
to emulate real tissue mechanical properties. Therefore, our first approach was to
explore printer Stratasys J750™ multi-material capabilities. A matching between
shore A hardness scale values of printing available materials and Knee Joint

16
anatomical structures were made. A model of ACL-R proposed in the previous
study without considering the designed fiber matrix was printed after materials
matching (Table 3). The materials Digital ABS™, Agilus30™, and Tango:
FLX950™-FLX930™ (rubber-like materials family) were selected for printing
bone components and soft tissues respectively. High Mix 27µm layer thickness
mode was set for the model printing.
Table 3: Proposed material matching.

Model Structures Materials selected Durometer selected (Shore A)

Bones Digital ABS™ A 95

Ligaments Agilus30 FLX2040™ A 35
Tendons Agilus30 FLX2040™ A 35
Retinacula Agilus30 FLX2040™ A 30
Menisci TangoGrayFLX950™ A 75
Cartilage TangoFLX930™ A 28

Unfortunately, the printing trial was not satisfactory, ligaments and tendons failed
easily after simulating flexo-extension movements (Figure 8.)

Figure 8: Printing trial scale 1:2. ACL-R model without collagen fibers matrix
structure. Knee Joint lateral views in flexion with BPTB graft replacing ACL,
rupture of the LPL, and MCL.
After examining the results of the first printing trial, for the printing materials
selection of the proposed models, we took advantage of the Stratasys J750 ™
printer's capability. Different combinations of hardness values of fibers were
proposed. All models are based on the Knee Joint model. The Mechanical

17
characterization of the proposed combinations was conducted in Northwestern
University Kaiser Lab using an Instron S3300 (Canton, MA) uniaxial testing
instrument, following test designation D412-C for rubbers and elastomers. To
perform the uniaxial tensile tests, it was necessary to design bone specimens with
fibers inside (Table 4), in SolidWorks software following standard specifications
(Figure 9.). Three bone specimens (n=3) were printed for each combination.
Dimensions (thickness, length, width) of each bone specimen were measured with
calibrator, values were set in the software BlueHill, Instrom SA (France, Elancourt).
The tensile test was set using a test speed of 10 mm/min. Each bone specimen was
attached between the materials testing system extensometer grips to apply tensile
loads. The test was performed until the bone specimen failed. Data was recorded
and exported to Microsoft Excel. The procedure was repeated with all bone
specimens. Strain-stress curves of each specimen were elaborated in Matlab
(MathWorks R2018). From them, average curves were created for each
combination, elastic modulus, yield strength, and proportional limit values were
reported in the results section.
Table 4: Fibers-specimen hardness combinations (Agilus30) for the tensile test.

Material Fibers Durometer (Shore A) Specimen Durometer (Shore A)

combinations

No 1 A.50 A60
No 2 A.50 A55
No 3 A.50 A40
No 4 A.60 A70

18
Figure 9: Flowchart application standard D412-C. (A) Dimensions bone specimen
(mm), (B) Render bone specimen in SolidWorks software, (C) Printed bone
samples with different combinations, (D) Bone specimen attached to a materials
testing system Instron S3300 uni-axial testing instrument, to apply tensile load.

Results
Mechanical test.
The tensile test was performed to make a stiffness comparison (directly related to
the slope of the linear region) of the different material combinations and Knee Joint
soft tissues. We performed three mechanical tests for each combination to obtain
average values reported. The approximate elastic modulus and linear tendencies
were calculated in the elastic region (a linear approximation from endpoints of the
toe-region to the yield strength point) of the stress-strain curves. The Table 5.
includes values of Average-standard deviation of proportional limit (AVG/STD-
PL), Average-standard deviation of yield strength (AVG/STD-YS), Average-
standard deviation of elastic modulus (AVG/STD-E) and linear adjustment
coefficient or Pearson's correlation coefficient R2, which varies between 0.980-
0.991.

19
Figure 10: Experimental average Stress-Strain curves of the Knee Joint soft tissues
and material combinations.
20
Figure 11: Average linear tendencies for Knee Joint soft tissues and material
combinations.
21
Table 5: Tensile test data. Strain-stress curve properties of different printing
materials combinations.

Material AVG- STD-PL AVG- STD-YS AVG- STD-E 𝑅2

combinations PL YS E
(MPa) (MPa) (MPa)

No 1 1.533 0.030 1.560 0.008 0.769 0.011 0.980

No 2 1.436 0.035 1.516 0.034 0.751 0.009 0.975
No 3 1.486 0.012 1.580 0.008 0.660 0.005 0.970
No 4 2.336 0.016 2.426 0.002 1.822 0.003 0.991

We compared soft tissue curves (45) with the curves obtained from different
proposed combinations; both show the same mechanical pattern exhibiting an
approximate linear behavior (Figure 11). When we were comparing the stiffness
and elastic modulus values, we found that combinations No 1-4 were the most
similar to real structures with elastic modules of 1.8 and 0.7 respectively (Table 6).
Table 6: Matching of the soft tissues with the selected combinations for the PSAMs
models.

Knee Joint soft Elastic modulus Selected combination Elastic modulus

tissues (MPa) (MPa)

MCL 3.23 No 4 1.8

PT-QT 3.5 No 4 1.8
MPL-LPL 2.41 No 1 0.76
PCL 2.36 No 1 0.76
LCL 2.29 No 1 0.76
ACL 1.93 No 1 0.76

Agilus30: FLX2040™ printing materials in the combinations No 1 (A50 fibers and

A60 body) - 4 (A60 fibers and A70 body) were chosen to print ligaments, tendons,
and retinacula according to the matching. Tango FLX930™ (A28) - FLX950™
(A75) printing materials were chosen to print cartilage surfaces and menisci
respectively. Finally, Digital ABS™ (A95) was chosen to print bone components.

22
3D Printing of the Patient-specific anatomical models.
Three patient-specific Knee Joint models ACL-R preview study (34), ACL-R, and
TKA were printed. The models produced were accurate with no difference in size
and positioning relative to planed 3D computer-generated PSAMs replicating fibers
and realistic features of the Knee Joint anatomy. In this study, we did not evaluate
the anisotropy of PolyJet AM technology (two critical factors involved orientation
and LT). We have set PSAMs printing using High Mix six different materials
(0.027mm LT) mode according to factory default settings. We located the model
perpendicular to the printing tray in which, Z-axis of the SolidWorks GRS coincides
with the Z-axis of the printing tray. The multi-material 3D printing enabled the
combination of hard and elastic materials in a single project, enabling the 3D
printing of the hierarchical structure, custom-made SG, and PSP for TKA. After the
completion of printing, wax-like support material was removed using a pressure
water gun, the whole process took about 5 minutes. One of the advantages of the
use of PolyJet AM technology is the easy support material removal. Indeed, other
technologies such as FDM use soluble plastics in chemical baths at high
temperatures as a method of remove support material. The removal process can take
approximately a day or more, which extends the manufacturing chain. In general,
the printing price of PSAMs could be considered profitable about the costs
associated with the use of cadaver models or traditional educational models. This
was USD 570 with a production time of 28 hours. Finally, it was possible to validate
the collagen fibers matrix structure designed and material proposed combinations
in the models. After removing support material, the PSAMs were evaluated
manually simulating 50 flexo-extension cycles using a three-in-one multi-purpose
oil (ACL-R model was tested without BPTB positioned). The PSAMs did not
present rupture or wear in their connective structures or at the insertion points. The
custom-made SG matches perfectly with the anatomy of the PT in the ACL-R
model. Likewise, as it was planned, the BPTB measurements match perfectly with
the SG dimensions. The custom-made implants of the TKA model accomplished its
requirements and established preliminary planning but the range of motion was
more limited.

23
 Femoral tunnel

Fibers inside MCL

BPTB graft

Fibers inside PCL

24
Surgical guide

Fibers

Femoral tunnel

Tibial tunnel

25
 Tibial and
patellar bearings

Femoral component

Tibial component Tibial component

Figure 12: Illustration of Knee Joint PSAMs: (1) Anterior-Lateral- Posterior views of ACL-R (preview study), (2) ACL-R, and (3)TKA models. Final
models showed the same details as the 3D computer-generated PSAMs, where the fibers were evident in the grafts and other anatomical structures.

26
Discussion
The aim of the current study was to take advantage of the multi-color and multi-
material Stratasys J750™, and to use CAD software to mimic anatomical features
of the soft tissue structures. In addition, the SG and PSP were designed to improve
surgical outcomes in orthopedics, particularly in ACL-R as well as providing
functional models for TKA. We concluded that the approach mimicking the
hierarchical structure of the Knee Joint connective soft tissues was successful.
When we compared Stress-Strain curves for Knee Joint soft tissues and material
combinations as well as linear tendencies (Figures 10-11), we highlighted that
mechanical patterns and stiffness are comparable, despite different elastic modulus
values. For the proposed combinations, Pearson correlation coefficients 𝑅2 , were
close to one, which means that data adjustment was a good approximation of the
linear region, representative of the elastic behavior of ligaments and tendons (Table
6). From a functional point of view, when we compared the initial ACL-R model
(without hierarchical structure) and ACL-R proposed model, we confirmed that the
last one achieved expected results without connective structures wear or rupture.
In general, the proposed PSAMs withstood repetitive flexo-extension cycles
without problems (Figure 12).
The custom-made ACL printed surgical guide matches perfectly with the anatomy
of the PT in the ACL-R model. This would enable the surgeon to solve the problem
related to graft tunnel length mismatch and uncertainty in the graft harvest,
ensuring dimensions established in the pre-operative plan. The main limitation of
this approach is surgical validation, it is expected that the SG can be used in the
application of real cases of ACL-R. Figure 13. Shows the comparison between the
traditional approach and the proposed solution.

27
Figure 13: Traditional approach for BPTB autograft harvest using a ruler to
measure graft dimension and proposed approach using ACL surgical guide.
Finally, in our study, the purpose of the TKA model was to fulfill the functional
requirements established in preoperative planning. There were two main
limitations: the PSP designed considered an extreme case of OA in which the
cruciate ligaments are sacrificed, which is not recommended from the medical
criteria. Besides, the approach was based on a Knee joint under normal conditions
and it does not represent a real OA condition, therefore despite the results, the PSP
designed are not a real application for TKA. Other critical factors determine the
success or failure of TKA that this study does not consider.
Our study serves as the first step and a proof of concept for the accurate creation
of PSAMs, SG, and PSP.

Conclusion
The full-color, multi-material 3D printer: Stratasys J750™ has demonstrated its
capability to produce functional PSAMs with high fidelity, cost, and short-duration
from planning to manufacturing. The proposed PSAMs offer a diverse range of
applications. These can be considered as an alternative to replacing cadaver
specimens for medical training, pre-operative planning, research and education
purposes, and predictive models validation. We highlight the potential of PolyJet
AM technology combined with specialized medical software as a path to change
the way specialists and researchers plan, execute, and validate complex procedures.

Abbreviations
AM: additive manufacturing, 3DP: three-dimensional printing, PSAMs: Patient-
specific anatomical models, DICOM: digital imaging and communications in
medicine, CT: computed tomography, MRI: magnetic resonance imaging, US:
ultrasound, CAD: computer-aided design, CAE: computer-aided engineering ,
TKA: total knee arthroplasty, PSP: patient-specific prostheses, PSI: patient-
specific instrumentation, ACL: anterior cruciate ligament, ACL-R: anterior
cruciate ligament reconstruction, SG: surgical guide OA: osteoarthritis, BPTB:
bone-patellar tendon-bone, STG: semitendinosus-gracilis, CS: cruciate sacrificing,
UV: ultraviolet, LT: layer thickness, ABS: acrylonitrile butadiene styrene, FDM:
Fused Deposition Modeling, STL: standard triangle language, PF: patello-femoral,

28
TF: tibio-femoral, PT; patella tendon, QT: quadricep tendon, GRS: global
reference system, PCL: posterior cruciate ligament, MCL: medial collateral
ligament, LCL: lateral collateral ligament, MPL: medial patella-femoral ligament,
LPL: lateral patella-femoral ligament, MPR: medial patellar retinacula, LPR:
lateral patellar retinacula, TT: transtibial technique, FMAa: anatomic-mechanic
femoral angle, FDLAa: anatomic lateral distal femoral angle, FDLMa: mechanical
lateral distal femoral angle, AVG/STD-PL: average-standard deviation of
proportional limit, AVG/STD-YS: average-standard deviation of yield strength,
AVG/STD-E: average-standard deviation of elastic modulus.

Acknowledgments
We would like to express our sincere gratitude to the Engineer. Eric Renteria,
application Engineer at Materialise US. Who provided us a provisional license to
try out Mimics innovation suite. The specialized software Materialise 3-Matic
research version proved quite useful and versatile. Its measurement, simulation,
positioning, meshing, and design tools enabled the development of the proposed
models. Likewise, our sincere thanks go to Engineers: Kunal Shah, a research
engineer at the Center for Bionic Medicine within the Shirley Ryan AbilityLab and
Samuel Sung, a research assistant at Northwestern University KaiserLab who
provided me support and access to the laboratories and research facilities. Without
their continued support, it would not be possible to conduct this research.

Authors’ contributions
The author(s) read and approved the final manuscript.

Authors’ information
Not applicable.

Funding
The Shirley Ryan Abilitylab and Northwestern University supported this study.

Availability of data and materials

The 3D computer-generated PSAMs used to support the findings of this study are
available from the corresponding author on reasonable request.

29
Ethics approval and consent to participate
Not applicable.

Consent for publication

Not applicable.

Competing interests
The corresponding authors declare that they have no competing interests related to
this study.

References
1. Ashish, Ahmad N, Gopinath P, Vinogradov A. 3D Printing in Medicine
In: 3D Printing Technology in Nanomedicine. Elsevier; 2019. p. 1–22. DOI:
10.1016/B978-0-12-815890-6.00001-3
2. Tood P. The impact of implementing 3D Printing at the point-of-care In:
Materialise medical. Whitepapers. 2018.
https://www.materialise.com/en/resources/medical/whitepapers/impact-of-
implementing-3d-printing-at-point-of-care
3. Yan Q, Dong H, Su J, Han J, Song B, Wei Q, et al. A Review of 3D
Printing Technology for Medical Applications. Engineering. 2018;4(5):729–42.
DOI: 10.1016/j.eng.2018.07.021
4. Anderson PA. 3D Printing for Education and Surgical Planning in
Orthopedic Surgery. 3D Printing in Orthopaedic Surgery. Elsevier Inc.; 2019. 55–
63 p. DOI: 10.1016/B978-0-323-58118-9.00006-3
5. Bücking TM, Hill ER, Robertson JL, Maneas E, Plumb AA, Nikitichev
DI. From medical imaging data to 3D printed anatomical models. PLoS One.
2017;12(5):1–10. DOI: 10.1371/journal.pone.0178540
6. Lal H, Patralekh MK. 3D printing and its applications in orthopaedic
trauma: A technological marvel. J Clin Orthop Trauma. 2018;9(3):260–8. DOI:
10.1016/j.jcot.2018.07.022
7. Nadeem M, Jackson E, Rachala SR. 3D Printing in Hip and Knee
Arthroplasty. 3D Printing in Orthopaedic Surgery. Elsevier Inc.; 2019. 171–177 p.
DOI: 10.1016/B978-0-323-58118-9.00014-2

30
8. Joskowicz L, Hazan EJ. Computer-aided orthopedic surgery: Incremental
shift or paradigm change? Adv Exp Med Biol. 2018;1093:21–30. DOI:
10.1007/978-981-13-1396-7_2
9. Kim JW, Lee Y, Seo J, Park JH, Seo YM, Kim SS, et al. Clinical
experience with three-dimensional printing techniques in orthopedic trauma. J
Orthop Sci. 2018;23(2):383–8. DOI: 10.1016/j.jos.2017.12.010
10. Vaishya R, Vijay V, Vaish A, Agarwal A. Three-dimensional printing for
complex orthopedic cases and trauma: A blessing. Vol. 15, Apollo Medicine. 2018.
p. 51–4. DOI: 10.4103/am.am_51_18
11. Kenneth R. Foster. Philosophy and Engineering. In: Michelfelder DP,
Newberry B, Zhu Q, editors. Cham: Springer International Publishing; 2017. p.
211–28. DOI: 10.1007/978-3-319-45193-0_16
12. Matsumoto JS, Morris JM, Rose PS. 3-Dimensional printed anatomic
models as planning aids in complex oncology surgery. JAMA Oncol.
2016;2(9):1121–2. DOI: 10.1001/jamaoncol.2016.2469
13. Jones GG, Clarke S, Jaere M, Cobb J. 3D printing and unicompartmental
knee arthroplasty. EFORT Open Rev. 2018;3(5):248–53. DOI: 10.1302 / 2058-
5241.3.180001
14. Qiu B, Liu F, Tang B, Deng B, Liu F, Zhu W, et al. Clinical Study of 3D
Imaging and 3D Printing Technique for Patient-Specific Instrumentation in Total
Knee Arthroplasty. J Knee Surg. 2017;30(8):822–8. DOI: 10.1055/s-0036-
1597980
15. Soni A, Kumar Modi Y, Agrawal S. Computed tomography based 3D
modeling and analysis of human knee joint. Mater Today Proc. 2018;5(11):24194–
201. DOI: 10.1016/j.matpr.2018.10.214
16. Crawford SN, Waterman BR, Lubowitz JH. Long-term failure of anterior
cruciate ligament reconstruction. Arthrosc - J Arthrosc Relat Surg.
2013;29(9):1566–71. DOI: 10.1016/j.arthro.2013.04.014
17. Olsson O, Isacsson A, Englund M, Frobell RB. Epidemiology of intra-
and peri-articular structural injuries in traumatic knee joint hemarthrosis – data
from 1145 consecutive knees with subacute MRI. Osteoarthr Cartil.
2016;24(11):1890–7. DOI: 10.1016/j.joca.2016.06.006

31
18. Rankin I, Rehman H, Frame M. 3D-Printed Patient-Specific ACL
Femoral Tunnel Guide from MRI. Open Orthop J. 2018;12(1):59–68. DOI:
10.2174/1874325001812010059
19. Ni J, Li D, Mao M, Dang X, Wang K, He J, et al. A Method of Accurate
Bone Tunnel Placement for Anterior Cruciate Ligament Reconstruction Based on
3-Dimensional Printing Technology: A Cadaveric Study. Arthrosc - J Arthrosc
Relat Surg. 2018;34(2):546–56. DOI: 10.1016/j.arthro.2017.08.288
20. Tisherman R, De Groot J, Rothrauff B, Byrne K, Meredith SJ, Musahl V.
The role of anatomic ACL reconstruction in ACL revision surgery. Ann Jt.
2018;3(3):103–103. DOI: 10.21037/aoj.2018.11.14
21. Samuelsson K, Andersson D, Karlsson J. Treatment of Anterior Cruciate
Ligament Injuries With Special Reference to Graft Type and Surgical Technique:
An Assessment of Randomized Controlled Trials. Arthrosc - J Arthrosc Relat Surg.
2009;25(10):1139–74. DOI: 10.1016/j.arthro.2009.07.021
22. Gans I, Retzky JS, Jones LC, Tanaka MJ. Epidemiology of Recurrent
Anterior Cruciate Ligament Injuries in National Collegiate Athletic Association
Sports: The Injury Surveillance Program, 2004-2014. Orthop J Sport Med.
2018;6(6):1–7. DOI: 10.1177/2325967118777823
23. Kaeding CC, Pedroza AD, Reinke EK, Huston LJ, Spindler KP,
Amendola A, et al. Risk Factors and Predictors of Subsequent ACL Injury in Either
Knee After ACL Reconstruction: Prospective Analysis of 2488 Primary ACL
Reconstructions from the MOON Cohort. Am J Sports Med. 2015 Jul
21;43(7):1583–90. DOI: 10.1177/0363546515578836
24. Whitehead TS. Failure of Anterior Cruciate Ligament Reconstruction.
Clin Sports Med. 2013 Jan;32(1):177–204. DOI: 10.1016/j.csm.2012.08.015
25. Vaishya R, Okwuchukwu MC, Agarwal AK, Vijay V. Does Anterior
Cruciate Ligament Reconstruction prevent or initiate Knee Osteoarthritis? -A
critical review. J Arthrosc Jt Surg. 2019;6(3):133–6. DOI:
10.1016/j.jajs.2019.07.001
26. Friel NA, Chu CR. The Role of ACL Injury in the Development of
Posttraumatic Knee Osteoarthritis. Clin Sports Med. 2013;32(1):1–12. DOI:
10.1016/j.csm.2012.08.017
27. Huang W, Ong TY, Fu SC, Yung SH. Prevalence of patellofemoral joint
osteoarthritis after anterior cruciate ligament injury and associated risk factors: A

32
systematic review. J Orthop Transl. 2020;22:14–25. DOI:
10.1016/j.jot.2019.07.004
28. Van Eck CF, Kropf EJ, Romanowski JR, Lesniak BP, Tranovich MJ, van
Dijk CN, et al. ACL graft re-rupture after double-bundle reconstruction: Factors
that influence the intra-articular pattern of injury. Knee Surgery, Sport Traumatol
Arthrosc. 2011;19(3):340–6. DOI: 10.1007/s00167-010-1297-8
29. Kyritsis P, Bahr R, Landreau P, Miladi R, Witvrouw E. Likelihood of
ACL graft rupture: Not meeting six clinical discharge criteria before return to sport
is associated with a four times greater risk of rupture. Br J Sports Med.
2016;50(15):946–51. DOI: 10.1136/bjsports-2015-095908
30. Wada O, Gamada K, Aoyama N, Mizuno K, Iwasaki Y. A difference in
rotational alignment of the tibio-femoral joint after anterior cruciate ligament
reconstruction between the bone-patellar tendon-bone and semitendinosus-gracilis
grafts. Clin Biomech. 2019;65:45–50. DOI: 10.1016/j.clinbiomech.2019.03.015
31. Gupta PK, Acharya A, Mourya A, Mahajan P. Comparison of patellar
tendon versus hamstrings autografts for anterior cruciate ligament reconstruction
in Indian population: A randomised control trial study. J Clin Orthop Trauma.
2019;10(3):581–5. DOI: 10.1016/j.jcot.2018.04.018
32. Duchman KR, Lynch TS, Spindler KP. Graft Selection in Anterior
Cruciate Ligament Surgery: Who gets What and Why? Clin Sports Med.
2017;36(1):25–33. DOI: 10.1016/j.csm.2016.08.013
33. Li RT, Lorenz S, Xu Y, Harner CD, Fu FH, Irrgang JJ. Predictors of
radiographic knee osteoarthritis after anterior cruciate ligament reconstruction. Am
J Sports Med. 2011;39(12):2595–603. DOI: 10.1177/0363546511424720
34. Salehghaffari S, Dhaher YY. A model of articular cruciate ligament
reconstructive surgery: A validation construct and computational insights. J
Biomech. 2014;47(7):1609–17. DOI: 10.1016/j.jbiomech.2014.03.003
35. Lizaur-Utrilla A, Martinez-Mendez D, Gonzalez-Parreño S, Marco-
Gomez L, Miralles Muñoz FA, Lopez-Prats FA. Total Knee Arthroplasty in
Patients With Prior Anterior Cruciate Ligament Reconstruction. J Arthroplasty.
2018;33(7):2141–5. DOI: 10.1016/j.arth.2018.02.054
36. Chen D, Shen J, Zhao W, Wang T, Han L, Hamilton JL, et al.
Osteoarthritis: Toward a comprehensive understanding of pathological
mechanism. Bone Res. 2017;5. DOI: 10.1038/boneres.2016.44

33
37. Liddle AD, Pegg EC, Pandit H. Knee replacement for osteoarthritis.
Maturitas [Internet]. 2013;75(2):131–6. DOI: 10.1016/j.maturitas.2013.03.005
38. Sharkey PF, Hozack WJ, Rothman RH, Shastri S, Jacoby SM. Why are
total knee arthroplasties failing today? Clin Orthop Relat Res. 2002;(404):7–13.
DOI: 10.1097/00003086-200211000-00003
39. Gay P, Blanco D, Pelayo F, Noriega A, Fernández P. Analysis of Factors
Influencing the Mechanical Properties of Flat PolyJet Manufactured Parts.
Procedia Eng. 2015;132:70–7. DOI: 10.1016/j.proeng.2015.12.481
40. Biewener AA. Tendons and ligaments: Structure, mechanical behavior
and biological function. Collagen Struct Mech. 2008;269–84. DOI: 10.1007/978-
0-387-73906-9_10
41. Provenzano PP, Vanderby R. Collagen fibril morphology and
organization: Implications for force transmission in ligament and tendon. Matrix
Biol. 2006;25(2):71–84. DOI: 10.1016/j.matbio.2005.09.005
42. Hsieh AH, Tsai CM-H, Ma Q-J, Lin T, Banes AJ, Villarreal FJ, et al.
Time-dependent increases in type-III collagen gene expression in medial collateral
ligament fibroblasts under cyclic strains. J Orthop Res. 2000;18(2):220–7. DOI:
10.1002/jor.1100180209
43. Grawe B, Smerina A, Allen A. Avoiding graft-tunnel length mismatch in
anterior cruciate ligament reconstruction: The single-bone plug technique.
Arthrosc Tech. 2014;3(3):e417–20. DOI: 10.1016/j.eats.2014.04.003
44. Jingjit W, Poomcharoen P, Limmahakhun S, Klunklin K, Leerapun T,
Rojanasthien S. Femoral mechanical-anatomical angle of osteoarthritic knees. J
Med Assoc Thail. 2014;97(12):1314–8.
45. Mesfar W, Shirazi-Adl A. Biomechanics of the knee joint in flexion under
various quadriceps forces. Knee. 2005;12(6):424–34. DOI:
10.1016/j.knee.2005.03.004

34