Keywords
Hoffa's fracture - pseudarthrosis - printing, three-dimensional - orthopedic surgery
Introduction
In several fields of medicine, such as orthopedics, anatomical models and personalized
implants are manufactured for an accurate preoperative planning, the simulation of
surgeries, the training of staff, and better communication with the patient.[1]
[2]
[3]
[4]
Three-dimensional (3D) printing, also called rapid prototyping (RP) or additive manufacturing
(AM), enables the printing of models reproducing the actual anatomy. These can improve
the surgeon's anatomical understanding and interpretation. This technology enables
the use of personalized guides to optimize the surgical outcomes. Implants and arthroplasties
can be developed according to an individual's anatomy and can also help in the anatomical
assessment of the surgical treatment.[5]
[6]
Virtual 3D planning enables the surgeon to visualize and understand the complex anatomy
and to digitally plan a corrective osteotomy to restore the anatomy and normal function
or the best implant placement, for example.
The objective of the present study is to evaluate a proposed 3D printing process of
a biomodel developed with the aid of fused deposition modeling (FDM) based on the
computed tomography (CT) scans of an individual with nonunion of the coronal femoral
condyle fracture (Hoffa fracture).
Materials and Methods
To demonstrate the use of the AM technology in orthopedic surgery, we chose a clinical
case of an individual with a Hoffa fracture nonunion. We performed the study of bone
geometry, virtual preoperative planning, and preoperative planning with the printing
of a 3D anatomical model in full scale (1:1) of the distal region of the femur. In
this model, the fixation of the nonunion was simulated with osteosynthesis by plate
and screws. After the surgical treatment, the implant position was analyzed (in the
anatomical model and in the patient) through the measurement of the angles of the
implants oin relation to bone structure landmarks. These measurements enabled us to
analyze the reproducibility of the surgical planning and simulation. The present study
was performed after approval by the institutional Ethics Committee (CAAE: 94788418.2.0000.5547).
The patient is a 44-year-old male who, in november 2010, suffered a motorcycle accident
that resulted in severe polytrauma with an open segmental fracture of the left femur
(fracture of the proximal and distal regions). He presented sequelae of the trauma
to the left lower limb with nonunion and angular deformity.
After six years of the accident, he started presenting with progressive pain. The
X-ray and CT scans were performed, and a neglected Hoffa's fracture nonunion was diagnosed
(Letteneur type III) ([Fig. 1]).[7]
Fig. 1 Sagittal, axial, and coronal CT scans.
For the acquisition of bone images to print the anatomical model, we used the GE lightspeed
VCT (GE Healthcare, Chicago, IL, United States) CT scanner, manufactured in 2008,
with 64 channels. The scans were performed following a specific image acquisition
protocol for bone tissue established by the local radiology team, as shown in [Table 1].
Table 1
|
Parameters
|
Description
|
|
Field of view
|
17 × 17 cm
|
|
Scout
|
Standard protocol of the scanner
|
|
Region of interest
|
Knee
|
|
Voltage (KV)
|
120
|
|
Amperage (mA)
|
298
|
|
Pitch
|
512 × 512
|
|
Colimation
|
Large body
|
|
Slice thickness
|
0,6 mm
|
|
Slice increment
|
0,969 mm
|
To create the virtual anatomical model to be used, segmentation was performed (the
separation of bone tissue from the CT scans) using the Invesalius (Centro de Tecnologia
da Informação [CTI] Renato Acher, Campinas, SP, Brazil) software, version 3.1.1, as
shown in [Fig. 2]. The bone segmentation was performed by an automatic algorithm of the software that
identified bone tissue in the radiodensity threshold window between 226 and 2014 Hounsfield
units (HU). The program created a mask to identify the segmented tissue, which, in
this case, was green, as shown in [Fig. 2]. The software enabled the visualization of the reconstructed object in various positions
in space, helping to better understand the bone geometry. Volumetric reconstruction
of the object was performed, and a file was created in the standard triangle language
(STL) format, which can be exported to a computer-aided design (CAD) environment for
subsequent virtual modeling of the object or even its 3D printing.
Fig. 2 Screen image of the Invesalius software showing the bone segmentation. Note: The
figure shows the images of the axial, sagittal, coronal sections, as well as the image
of the reconstructed volume of the segmented bone tissue.
The object was modeled and rendered through the Meshmixer (Autodesk, San Rafael, CA,
United States) software, version 3.5, with the levelling of the internal and external
surface of the bone ([Fig. 3]).
Fig. 3 Virtual bone model before rendering with the Meshmixer software. Note: A mesh was
developed to better understand the irregularities and flaws on the surface of the
object to facilitate their subsequent correction.
The use of a software with CAD technology enables the development of the virtual surgical
planning (VSP) and the reduction of fragments ([Fig. 4]). The non-consolidated fragments were separated virtually to better study the lesionn,
followed by their reduction in the anatomical position, and the positioning of the
implants in the desired location. The entire virtual surgery process was performed
in a CAD environment using the Meshmixer software ([Fig. 4]).
Fig. 4 Images of the reduction of the nonunion fragments.
After the creation of the object in STL format, the file was exported to a G-code
generator; G-code is a language used by computer-aided manufacturing (CAM) software
to manufacture 3D objects. The main function of the G-code is to instruct the machine
to move geometrically in the three dimensions: X, Y, and Z. It is an extremely simple
and rudimentary language, consisting of a sequence pf lines of instructions in which
each line is responsible for a specific task, and the software is run line by line
until the end of the code. The G-code of the object was generated using the Slic3r
(developed by Alessandro Ranellucci) free software, version 1.3.0.
After the generation of the G-code, the information was exported to a CAM environment.
The Repetier-Host (HotWorld GmbH & Co. KG, Willich, Germany) free software, version
2.0.5, was used to set the printing features, such as the type of thermoplastic, polymer
density, filament diameter, extrusion temperature, printing speed, height (width)
of the layer, resolution, filling of the structure, and support of the parts.
The manufacturing of the printed model reproduced the internal and external profiles
of the cortical layer of the bone, which is completely filled with polymer, representing
the actual geometry of the bone ([Fig. 5]).
Fig. 5 Images of the 3D-printed bone model with the FDM technology in white ABS, showing
the distal femur and fragment of the lateral femoral condyle.
The surgical simulation was performed according to recent studies[7]
[8] that demonstrate that plate and screws generate more biomechanical stability in
the treatment of coronal fracture of the femoral condyle. A 5-hole locking compression
plate (LCP) with four 3.5 mm locked screws, a 4.0 mm cancelous screw, a 3.5 mm cortical
screw were used.
Fixation of the nonunion of the 3D anatomical model was performed according to the
principles of fixation of joint fractures developed by the Arbeitsgemeinschaft für
Osteosynthesefragen (AO) Foundation ([Fig. 6]).[9]
Fig. 6 Images of the 3D printed bone fixated with the LCP plate on the lateral portion of
the distal femur.
The plate was bent following the geometry of the lateral femoral condyle so that it
would be at 90° in relation to the nonunion line.The proper positioning of the implants
and the reduction of bone fragments were assessed through perioperative fluoroscopy
([Fig. 7]).
Fig. 7 Perioperative fluoroscopy images to assess the positioning of the implants.
After fixating the 3D anatomical model with the implants, stability of the fragment
and/or the plate and screws were assessed through manual displacement tests reproducing
the flexion and torsion forces.
The anatomical model was sterilized in ethylene oxide to be used during the surgical
procedure as a navigation guide ([Fig. 8]).
Fig. 8 Use of the 3D-printed anatomical model as a navigation guide for the anatomical landmarks
to place the implant according to the virtual planning.
Results
The 3D-printed anatomical model showed geometric and morphological characteristics
similar to those of the actual bone. The material used to print the model, the thermoplastic
polymer acrylonitrile butadiene styrene (ABS), provided the model with structural
strength that enabled the surgical simulation with the placement of the implants,
without causing breakage or loosening of the layers of the model. After inserting
the implants, the stability of the reduction and fixation of the fragments were assessed.
Using the described method, we were able to print a 3D anatomical model with the characteristics
of the actual object, which enabled the development of the VSP, a surgical simulation
to plan the best position of the implants and train the surgery team.
The assessment of the reproducibility of the technique, that is, determining if the
osteosynthesis performed on the patient's distal femur followed the virtual planning
and the simulation using the 3D-printed anatomical model, was based on the measurements
of the angles in comparison to the anatomical landmarks of the distal femur, the nonunion
line, and the position of the implant. Angle measurement was performed using the AnimatiPACS
(Animati netPACS, Santa Maria, RS, Brazil) software.
To measure the position of the implant, the anatomical landmarks were defined on anteroposterior
(AP) and Lateral view (L) radiographs, with the tangential tracing of the lines in
the anatomical landmarks to form angles with the implants or the nonunion line. On
the AP view, a line was defined as a tangential landmark to the distal end of the
femoral condyles. On the Lateral view, a line was defined as a tangential landmark
to the posterior cortex of the distal femoral metaphysis. ([Fig. 9]).
Fig. 9 Anatomical references on the AP and Lateral view radiographs with the tangential
tracing of lines on the landmarks and the formation of angles to measure the position
of the implants.
The radiographs (on AP and Lateral views) of the patient's knee and the 3D-printed
anatomical model show that the position of the plate and screws in relation to the
defined anatomical landmarks (articular surface of the femoral condyles and the posterior
cortical layer of the distal metaphysis of the femur) are different ([Table 2]).
Table 2
|
X-ray view
|
Knee
|
Anatomical model
|
|
Nonunion line
|
Anteroposterior
|
41.2°
|
38,1°
|
|
Lateral
|
43.8°
|
61,5°
|
|
Plate
|
Anteroposterior
|
49,6°
|
71,8°
|
|
Lateral
|
36,9°
|
26,0°
|
|
Proximal screw
|
Anteroposterior
|
57,2°
|
74,6°
|
|
Lateral
|
57,0°
|
32,1°
|
|
Distal screw
|
Anteroposterior
|
52,5°
|
62,8°
|
|
Lateral
|
50,9°
|
37,3°
|
The patient's knee radiographs on Lateral view and the printed anatomical model show
a proximity of the values of the angles between the plate and screws in relation to
the nonunion line ([Table 3]). The differences in percentage regarding the positioning of the implants in the
patient's knee in relation to the anatomical model were of 3.5% on the plate, 18%
on the proximal screw, and 4% on the distal screw.
Table 3
|
X-ray view
|
Nonunion line
|
|
Knee
|
Anatomical model
|
|
Plate
|
Lateral
|
93.0°
|
89.8°
|
|
Proximal screw
|
Lateral
|
70.2°
|
85.7°
|
|
Distal screw
|
Lateral
|
76.1°
|
79.4°
|
Discussion
Before the advent of computer systems to visualize digitized images, such as the picture
archiving and communication system (PACS), CT and magnetic resonance imaging (MRI)
scans were printed on film, with the loss of valuable information in the process.
The PACS changed the way of analyzing the images, providing another dimension to their
interpretation, since it enables the dynamica visualization of the 3D object from
various angles. The volumetric reconstruction of the studied structure can be exported
to CAD software for modeling and rendering, which make it possible to print the object.[10]
According to several authors,[6]
[11]
[12]
[13]
[14] currently, the 3D printing process in the medical field (used for teaching, planning,
and surgical training, and the development of surgical guides, implants, and prostheses)
consists of the steps showed in [Fig. 10].
Fig. 10 3D printing process in medicine.
Kim et al.[15] reported their clinical experience with the use of 3D printing techniques in orthopedic
trauma, which enables a better understanding of the fracture and its anatomical relationships,
and may be applied in the preoperative planning, in medical education, and in surgical
training and simulation.
In the present study, the process used for the acquisition of CT scans and rendering
and modeling in a CAD environment to create a virtual bone model for 3D printing through
FDM technology was successful. According to the literature, the parameters in the
acquisition of CT scans, such as the width and number of slices, are determining factors
for the accuracy of the volumetric reconstruction. The width of the slices was smaller
than 1 mm, which generated enough data for a volumetric reconstruction that was very
close to that of the actual bone. Segmentation was performed automatically by a computer
algorithm with the identification of bone tissue in the threshold window between 226
HU and 2014 HU, which enabled an accurate segmentation, with little need for the removal
of artifacts and smoothing the surface during rendering and modeling.
The use of CAD software, such as Invesalius (used for bone segmentation) and Meshmixer
(used for modeling and rendering), enabled us to perform the anatomical reduction
of bone fragments during the VSP.
The use of 3D printing technology has been growing exponentially in several fields,
including orthopedic surgery, and it is relevant in creation of biomodels, surgical
tools (such as guides and templates), implants and prostheses.
The anatomical model may also facilitate the communication among the medical team
and the patient and their family, providing information about the type of surgical
treatment, promoting a better understanding of the clinical picture and of the posible
complications, as well as better adherence to the treatment, which contributes to
improve the doctor-patient relationship.[4]
[16]
[17]
[18]
[19] The participant in the present study was instructed on the severity and prognosis
of the joint injury. For this purpose, we used the printed anatomical model submitted
to the surgical simulation, with the nonunion fixed with the implants chosen.
Studies[4]
[12]
[17]
[20] show the use of anatomical models with AM technology in surgical training. A wide
variety of domains including simulation accuracy, anatomical similarity, training
in the use of surgical instruments, use printed models for training surgeons. Several
studies have demonstrated the effective application of 3D printing in medical education
in orthopedics,[4]
[11]
[20]
[21]
[22]
[23]
[24]
[25]
[26] mainly in surgical procedures in sites of complex anatomy.[6]
There are several CAD software that enable the performance of the VSP to better understand
the spatial geometry and anatomical relations. It is possible to program less invasive
surgical approaches and anticipate the reduction of bone fragments, simulating the
definitive osteosynthesis.[27]
[28] The printing of biomodels provides additional information when compared to conventional
imaging scans and increases the knowledge about the lesion.[5]
[11]
[22]
[29]
Some meta-analyses and systematic reviews[3]
[30] on preoperative planning and 3D printing-assisted surgery in orthopedic trauma suggest
that this method reduces the surgical time, the levels of intraoperative blood loss,
and the fluoroscopy time.
In the present study, the printing of the 3D anatomical model with the FDM technology
enabled us to develop a careful VSP, to plan the position of the implants, and train
the surgical team. Moreover, the anatomical model fixated with the implants was used
as a navigation guide during surgery.There was some technical difficulty in positioning
the anatomical model so it would correspond to the precise position of the pantient's
knee on the AP radiograph. Due to the lack of other anatomical structures adhered
or attached to it, the model did not have geometric and anatomical landmarks that
would enable the precise reproduction of specifc X-ray technique for the knee. Thus,
the angles reveal that the patient's knee and the anatomical model were not in the
same position at the time the X-ray was performed, as shown in [Table 2].
In order to achieve a stable osteosynthesis, the screws and the plate must be at an
angle close to 90° in relation to the nonunion line, according the AO techniques,[9] which was achieved in the present study, as shown in [Table 3].
Despite the difference in the angles of the radiograph of the patient's knee and those
of the anatomical model, the positioning of the implants was adequate, and we were
able to reproduce the technique of placing the plate and screws as in the simulation
in the anatomical model. This resulted on great fixation stability, and enabled a
satisfactory reduction and the anatomical reconstruction of the joint.
According to the radiographic analysis, great accuracy was verified in the reproducibility
of the VSP and in the surgical simulation compared with the outcome of the osteosynthesis
of the Hoffa's fracture nonunion.
Conclusion
The protocol suggested for the acquisition and segmentation of the CT scans of the
knee of a patient with Hoffa's fracture nonunion proved to be effective for the volumetric
reconstruction and rendering of the 3D anatomical model. The model enabled the performance
of a careful VSP, as well as the simulation of the ostheosyntesis. According to the
radiographic analysis, great accuracy was observed in the reproducibility of the surgical
simulation. The use of the 3D-printed model proved to be effective and useful in the
planning and surgical treatment of Hoffa's fracture nonunion.
