Preprocessing of Medical Image Data for Three-Dimensional Bioprinted Customized-Neural-Scaffolds

Abstract

Despite the promising nerve regeneration research, the lack of a therapeutic scaffold to completely restore neurological function is still ongoing and the demand in providing patients with a successful therapeutic outcome is ever increasing. With the advancements in three-dimensional (3D) bioprinting, custom scaffolds with a predetermined size and shape have been successfully fabricated, which can be utilized in nerve damage regenerative attempts to potentially aid in the regaining of nerve function. With the documented success of graphical processing unit (GPU) implementation utilized for image-guided surgeries of tubular and organ structures, we propose the implementation of known processing methods as a means to drastically decrease the time required to process medical images related to nerve damage. In addition, we further propose that the merging of medical image cropping and 3D printing techniques provides a novel approach for providing patient-specific customized-neural-scaffolds for patients suffering with newly acquired nerve damage. Finally, we provide a proposed schematic that incorporates the implementation of GPUs and 3D printing, which we propose will beneficially decrease the waiting times for medical staff to provide patients with customized neural treatments.

Impact Statement

Nerve damage, which can be devastating, triggers several biological cascades, which result in the insufficiencies of the human nervous system to provide complete nerve repair and regain of function. Since no therapeutic strategy exists to provide immediate attention and intervention to patients with newly acquired nerve damage, we propose a strategy in which accelerated medical image processing through graphical processing unit implementation and three-dimensional printing are combined to produce a time-efficient, patient-specific (custom-neural-scaffold) solution to nerve damage. This work aims to beneficially shorten the time required for medical decision-making so that improved patient outcomes are achieved.

Introduction

The nervous system plays an undeniable role in the functioning of several bodily systems, including, but not limited to, the coordination of locomotion, sensory perception, various aspects of homeostasis, and the propagation of information around the body.¹ Nerve system damage can be caused by physical trauma to the nerve fiber or through neurodegenerative disease mechanisms and affects both the peripheral nervous system (PNS) and the central nervous system (CNS).¹ In higher organisms with limited and variable nerve regeneration capabilities (largely attributed to the surrounding nerve environment), regeneration can be achieved (injury dependent) in the PNS, whereas regeneration in the CNS is limited due to a slower clearance of injury matter, the formation of an astroglial scar, and the impedance of neural impulses.^1–6 To restore neurological function, effective axonal communication must be established before, throughout, and after the injury gap.⁷ Despite a plethora of promising nerve regeneration studies, we are yet to achieve a successful therapeutic scaffold to overcome the poor regenerative abilities of the human nervous system and provide complete regain of neural functioning.^8–13

Owing to the fact that injury matter can lead to further nerve trauma, medical imaging has become a fundamental tool in the evaluation of sustained nerve damage.^14,15 Medical imaging modalities, namely magnetic resonance imaging (MRI) and computed tomography (CT) scans, are the most common imaging techniques in assessing patient nerve damage.^14,16 Medical imaging plays a vital role in, among others, evaluating preoperative and postoperative patient data, patient diagnosis and injury characterization, confirmation of the precise location of the nerve damage, assessment of the stability of the damaged area, rapid medical decision-making (including in-surgery decisions), and surgical planning, as well as assisting in avoiding further neurological deterioration.^14,16–19 Despite medical imaging providing invaluable assistance in the evaluation of patient nerve damage,^14,15,18,20 there are still pertinent challenges surgeons face during surgery. During surgery, injury and trauma to the nerve may cause abnormalities in the known nerve anatomy, and hence, the anatomy may vary from individual to individual. In addition, injury matter may enter the nerve fiber and further influence the nerve anatomy. In addition, the presence of water molecules along the nerve fiber makes it increasingly more difficult to ascertain the exact outline and orientation of the nerve fiber.²¹ Furthermore, despite the fact that medical imaging renders images in three-dimensional (3D), surgeries guided by images are limited as 3D images have to be viewed on flat screens and cannot provide a sufficient and complete understanding of the injury pathology.²² Therefore, standard imaging modalities alone cannot provide all the diagnostic evaluations required for all nerve injury cases.

As image-guided surgeries become more popular, the introduction of algorithmic image analysis has begun to overcome certain setbacks experienced by the standard imaging modalities. Regardless of the computational benefits, such requirements are time consuming and are not feasible for time-sensitive operations.²³ In response, the introduction of powerful graphical processing units (GPUs) has seen high-speed, more affordable, and more energy efficient medical image processing in comparison to central processing units (CPUs).²³ In addition, it is widely accepted that the entire medical dataset does not have to be computed, and therefore, the preprocessing of medical image data (image simplification²⁴ and visualization of the anatomical structures of interest²³) has become more popular in the clinical setting as a smaller dataset processing equates to faster processing times.²⁵ Moreover, parallel GPU architecture allows for commands to be run in parallel to substantially decrease possessing time; however, this is not an infinite feature as speedup is governed by the algorithm's sequential fraction.²³

The introduction of 3D bioprinting in various tissue engineering applications has greatly improved nerve regeneration attempts.²⁶ This is due to the fact that complex nerve constructs can be designed to closely mimic the native shape of the nerve requiring replacement.²⁶ Recent advancements in 3D printing technology has enabled researchers to fabricate scaffolds with defined dimensions, shapes, and pore structures to control properties for various tissue regeneration applications.^27,28 Therefore, the current advancements in 3D bioprinting has opened many doors with regard to customized patient treatment.^22,29,30 In addition, an article by Rengier et al.²² reviewed the use of medical imaging to 3D print constructs to improve patient care and surgical planning and aid in the design of customized tissues and implants. In terms of 3D bioprinting technologies, four main bioprinting techniques have emerged, namely stereolithography, laser-assisted 3D bioprinting, extrusion 3D bioprinting, and inkjet 3D bioprinting.^31–33 Each technology satisfies different criteria during the printing process and depending on the requirements of the fabricated structure, the appropriate bioprinting method will be selected.^31,33 A brief comparison between these technologies is presented in Table 1. Of the various 3D printing methods, extrusion-based 3D bioprinting is the most common technique used in tissue regeneration and regenerative medicine.³⁴ Extrusion-based 3D bioprinting is preferred for many reasons, including its capacity to fabricate constructs using multiple bioink types,^34,35 ease of operation, and scalability,³⁵ while being a relatively low-cost production method.³⁴ Moreover, extrusion-based 3D bioprinting utilizes a computer software program, for example, computer-aided design (CAD), to design and/or load medical image files (CT and MRI scans)²⁷ to print the intended scaffold.³⁵

Table 1.

A Brief Comparison of the Four Most Common Bioprinting Methods

Parameters	Stereolithography	Laser-assisted 3D bioprinting	Extrusion bioprinting	Inkjet 3D bioprinting
Technique summary	An additive manufacturing technique utilizing light to photocrosslink a photocurable bioink in a layer-by-layer manner^{32,33,36–39}	Technique involves the deposition and transfer of bioink/cells either using a pulsed laser or in a laser beam^{31–33,38,40}	Technique involves the dispensing of a viscous bioink through a nozzle using pneumatic or mechanical forces^{31,33,38,40,41}	A technique involving a noncontact approach using piezoelectric, thermal, or electromagnetic forces to eject consecutive drops of a bioink onto a chosen substrate^33,38,40,41
Viscosity	No limitation^39,41,42	1–300 mPa/s^32,41,42	30 mPa/s–>6 × 10⁷ mPa/s^32,37,41,42	3.5–12 mPa/s^32,37,41,42
Cell density	Medium <10⁸ cells/mL⁴¹	Medium <10⁸ cells/mL^32,41,42	High^{32,37,40–42}	Low <10⁶ cells/mL^32,37,41,42
Cell viability	>85%^32,37,41	>95%^33,39,41	40–80%^37,40	>85%^32,37,40
Resolution	High^{37,38,41–43}	High^37,38,41,42	Moderate^37,41,42	High^{33,37,38,40–42}
Print speed	Fast^32,41,42	Medium^32,38,41,42	Slow^{32,37,38,41,42}	Fast^{33,37,38,40–42}
Cost	Low^32,37,41,42	High^{32,33,37,38,40–42}	Medium^{32,37,40–42}	Low^{32,33,37,40–42}
Vertical printing capabilities	Good^32,37	Medium to low^32,37,38	Good^32,37	Poor^32,37
Advantages	Lower shear pressures due to nozzle-free approach³³	Clogging is avoided due to nozzle-free approach, deposition of bioink in solid or liquid phase^31,32,38	Can control scaffold porosity, ability to bioprint with high cell densities^32,33,40	Precise cellular positioning, can print low viscosity bioinks^32,33
Disadvantages	Requires intense UV curing (possible cell and DNA damage), possible cytotoxicity from photoinitiators, lengthy postprocessing times, and few compatible materials^33,36	Possible cell and DNA damage from laser exposure, low scalability and suitability^31–33	Technique is sensitive to bioink viscosity—if incorrect leakage can occur, printing pressures could potentially alter cellular morphology, shape, and function^33,38	Potential for higher cellular death due to higher nozzle temperatures, nozzles prone to clogging, continuous flow not sustained^32,33,38

3D, three dimensional.

Irrespective of the benefits introduced by 3D bioprinting, there are limitations with regard to neural 3D bioprinting across all 3D bioprinting method techniques. The biggest challenge with regard to neural 3D bioprinting is the lack of suitable bioinks with adequate biomimetic properties,^44,45 for example, the native electrical conductivity of the nerve fiber.⁴⁵ In addition, factors, including, but not limited to, bioink viscosity, bioink cytocompatibility, ability of the bioink to mimic the native extracellular matrix, and the capability of the final scaffold to retain structural integrity, are all aspects that have limited neural tissue 3D bioprinting.^44–46 Finally, under current frameworks, customized 3D printed constructs face notable regulatory challenges with regard to software system chain control and the role customized constructs play in the variation in performance of constructs postsurgery.³⁰

Despite the regulatory challenges regarding customized 3D printed constructs, the need for customized treatments is in high demand. Due to the nature of nerve damage, injuries are largely patient specific and current options are inadequate in repairing complex lesions. For this reason, personalized constructs can incorporate injury- and patient-specific considerations to customize designs on a per-patient basis. Therefore, customized patient treatments may be able to provide groundbreaking solutions to overcoming the current challenges in the repair of complex nerve injuries through the development of custom-neural-scaffolds (C-N-S).⁴⁷

Problem Statement

As previously mentioned, there is no known therapeutic scaffold that is able to completely restore neurological function. To compound the issue, nerve damage is particularly patient specific and nerve damage is complex as inflammatory events postinjury lead to the progressive loss of nervous tissue and therefore nerve functioning.⁴⁸ The schematic shown in Figure 1 provides a summary of the current approach in assisting and diagnosing newly acquired nerve damage in patients.^49–51 From the schematic, it is evident that the procedure is laborious (due to the wait period for diagnosis), while only providing palliative support to patients. While this is the best current option for patients, it is evident that a successful therapeutic scaffold with reduced waiting times is in high demand for patients living with newly sustained nerve damage. Interestingly, no study, to the best of our knowledge, has proposed the use of medical images coupled with GPU implementation and 3D bioprinting as a means to provide a solution to the current shortfalls in nerve regeneration.

FIG. 1.

Schematic displaying the current workflow used in the health care setting to diagnose potential nerve damage as well as the available treatment and management options.

Our Solution

Since no study has proposed the use of medical images and GPU implementation in an attempt to provide patient-specific 3D bioprinted C-N-S, this technical note proposes the implementation of data image processing algorithms as a means to drastically decrease the time required to process medical images related to nerve damage events. In addition, we propose that with GPU utilization and implementation from processing only the area of interest where nerve damage has occurred to the generation of the Gcode for 3D printing, we can achieve time-efficient data processing to potentially produce patient C-N-S. Finally, we envisage the development of an algorithm to provide a mechanism of examining the data (to ensure no computing errors have occurred) before the printing process.

Proposed Approach for GPU Implementation in Producing C-N-S

GPU implementation in medical imaging

As medical imaging instruments increase in sophistication, the sheer volume of data processing in this field has exploded to a point where processing time has become the limiting factor in further advancements.⁵² A modern-day solution to reducing processing time and increasing output speeds is with the implementation of parallel GPU systems.^52–54 The use of GPUs was incorporated originally into CT modalities as the reconstruction of CT data requires computationally intensive algorithms when compared to MRI data that require a more simple, fast inverse Fourier Transform image reconstruction.⁵⁴ Despite the benefits of GPU implementation, GPU memory is limited and may in fact limit the processing of large datasets.^23,25

The processing of preoperative data from differing imaging modalities has gained popularity in image-guided surgeries, such as in surgeries related to tubular structures (blood vessels and airways) and complete organ structures.^23,55 With this in mind, the processing of such data has to be completed as fast as possible for patients requiring surgery. The results from a study performed by Smistad et al. indicated that to achieve the desirable speed of processing the area of interest within the image, various algorithms performed on parallel GPU systems were shown to significantly reduce the time required to process preoperative medical image data.²³ This study highlighted that real-time processing of data is possible and, if utilized, could provide an interesting platform in providing the possible successful therapeutic outcomes required for nerve damage. To the best of our knowledge, processing of nerve bundles has been achieved using MRI images; however, GPU usage was not implemented in this study.⁵⁶

Use of medical imaging and the incorporation of GPUs in nerve regenerative attempts

Since MRI data have shown to have a higher resolution compared with CT scans in certain medical applications,^57,58 the extraction of nerve damage from MRI data is advisable. Despite this, it would be costly to implement the usage of MRI scanning throughout the health care sector. Therefore, this proposal is for the usage of images generated from both MRI and CT modalities. Due to the nature of nerve damage, we propose that successful segmentation and centerline extraction, utilizing known algorithms in the literature, can be employed to crop, process, extract, and reconstruct the images of the defined region where the damage has been sustained. This will help to reduce the time spent on the extraction and reconstruction of the damaged area of the nerve requiring replacement. It is further proposed that following image extraction and reconstruction, further algorithm development and GPU implementation can be applied to ensure quick slicing, data checking, and Gcode generation. Following this, the high-resolution data (from the medical images) will be sent to the 3D printer to print the patient-specific scaffolds (C-N-S). This could ultimately speed up the entire preprocessing of data from the capturing of the image to the printing aspect, which will have a direct benefit for patients with sustained nerve damage.

Overview of the Proposed Method

A review of the available literature indicated that there are numerous algorithms that have been developed and/or modified for the medial application at hand. As such, no one set of algorithms has been proposed to be the “gold standard” for all medical imaging processing requirements. Despite the number of algorithms proposed in the literature, this article will only consider investigations utilizing extraction and image segmentation in the processing of medical images for different anatomical regions of the body. In addition, this article will consider algorithms for both hollow tubular structures and solid organs as nervous tissues are rod-like structures with a dense bundling of axons and other associated structures.⁵⁹

Of the articles reviewed, despite the fact that different paths are selected to process the images, the outcome of the image processing is the same. That is, the generation of a segmented image of the region of interest. This section aims to provide an overview of the main steps to produce a segmented image that can be utilized for, among others, surgical planning, intraoperative data analysis, and diagnosis. Figure 2 provides an overview of the procedural proposal from the capturing of the image to the fabrication of the 3D bioprinted patient C-N-S.

FIG. 2.

Schematic providing an overview of the proposed procedure. The steps are as follows: the area of interest is captured, the image is processed and refined, a computer-aided design model is developed, and finally, the patient C-N-S is bioprinted. This image is adapted from Bücking et al.⁶⁰ under the Creative Commons Legal Code Attribution 4.0 International Licence. C-N-S, customized-neural-scaffold. Color images are available online.

Data preprocessing

Due to the fact that GPU memory is limited, the computing of entire medical image datasets may not be possible.^23,25 In addition, medical data contain large amounts of data volume that are computer intensive, time consuming, and unnecessary to compute if surgeons and medical staff are only interested in a defined region of the image.^25,60–62 According to Smistad et al.²⁵ anatomical areas considered to be outside of the region of interest are generally found at the borders of the captured image.²⁵ With this in mind, to ensure optimal GPU performance, with the employment of a cropping algorithm^25,61 or generic down sampling and quantization techniques,⁶² the volume of medical data required to be processed is beneficially minimized and in turn, execution time is advantageously reduced.^25,61 This in turn helps to ensure that the first stage in efficient data computation can occur timelessly before surgery. Due to the fact that cropping of datasets has been successfully achieved for airways,^25,61,63 blood vessels,^62,64,65 and coronary arteries,⁶⁶ as well as for the brain,²⁵ liver,^25,65,67 kidneys, spleen, aorta, gallbladder, pancreas, and vena cava (inferior),⁶⁷ we propose the need for a cropping algorithm to reduce the data size for medical images relating to newly acquired nerve damage.

In addition to data cropping, a smoothing data algorithm or a regulating algorithm needs to be utilized to reduce the dataset's sensitivity to noise, contrast, and size (Fig. 2).^25,61 Furthermore, smoothing will limit the chance of data leakage (data falling outside the area of interest).⁶¹ This is especially required for CT scan image data.^25,61 Due to the fact that cropping of datasets has been successfully achieved for the airways^25,61 and vessels,²⁵ we propose the usage of a smoothing algorithm to limit data leakage for nerve damage imaging.

Due to the fact that progressive loss of neural tissue occurs shortly after the primary injury, especially in spinal cord trauma,⁴⁸ we propose the implementation of the above-mentioned type of algorithms for nervous tissue image preprocessing. With time-efficient data preprocessing, we propose that decreasing the time required for the preprocessing image step will beneficially decrease the time required for the whole workflow and therefore can aid in quicker medical responses.

Centerline extraction and segmentation

Following image preprocessing, image centerline extraction and image segmentation need to be performed on the cropped and smoothed dataset. In brief, centerline extraction represents the area of interest at a structural level as a line is positioned to run through the center of the intended structure in the image.^23,68 Moreover, centerline extraction aids in describing 3D shapes exhibiting circular symmetry and is required for the analysis of shape and the processing of the given geometries.⁶⁹ In addition, by performing a centerline extraction, an accurate length of the structure can be obtained, which assists in the virtual planning and navigation of surgeries associated with tubular structures, for example, the aorta and coronary arteries.⁶⁹ Since the implementation of a GPU to perform parallel centerline algorithmic calculations, the previous limitations of poor accuracy and extensive computation times (CPU implementation) have been resolved.¹⁸ Due to the fact that centerline extraction has been successful in tubular structures, we envisage a similar algorithm implementation (after cropping and smoothing) working for nerve tissue medical image data, despite its compact nature.⁵⁹

Following centerline extraction, the modified dataset will then be optimized using an image segmentation algorithm. Since this proposal is for a decrease in GPU processing time, we propose the use of a semiautomated segmentation algorithm so that time efficiency and accuracy can be merged (automatic segmentation sacrifices accuracy⁶⁵ and manual segmentation is time consuming,⁶⁵ and in the development of the proposed C-N-S, neither are not be feasible). Principally, image segmentation is employed to relate individual elements with a common property (i.e., relating to a common organ).²³ Furthermore, image segmentation permits the visualization of the structure of interest, while removing additional information deemed unnecessary. In addition, image segmentation allows for the analysis of further features found within the structure of interest, for example, tumor volume calculations.²³ More importantly, segmentation allows for patient- and injury-specific considerations to be made.^23,60 In a article by Bücking et al.,⁶⁰ they provide an overview of freely available segmentation software that are applicable to any anatomical region of the body,⁶⁰ and because of this, such software needs to be considered in the production of patient C-N-S. Finally, the resulting image of the computation is a 3D rendering of the structure of interest (Fig. 2).⁶⁰ Due to the fact that anatomical anomalies such as aneurisms, stenoses, stents, and calcifications can also benefitted from such modeling,⁶² we see potential for image segmentation to aid in the detection of traumatic nerve damage as such modeling may assist in the detection of foreign matter in and around the injured nerve fiber.

3D bioprinting

The final stage in the formation of patient C-N-S is postprocessing and entails the conversion of the segmented image into a 3D bioprintable format. To achieve this, the segmented image needs to be converted into a 3D mesh (Fig. 2), using a CAD tool, so that it is in a form that is able to be bioprinted (Gcode). Postprocessing is required to repair any segmentation and staircasing errors, as well as removing additional structures that are unnecessary.⁶⁰ Furthermore, this implementation has seen the successful bioprinting of the lungs, ribs, and liver.⁶⁰ Due to the repair aspect of the postprocessing stage, we propose this method to be used for neural tissue medical imaging as due to the nature of nerve injuries, any checking mechanism will ensure the C-N-S will be as patient specific as it can be.

Patient Implications, Considerations, and Limitations of C-N-S

Due to the fact that the patient is the main priority in any clinical setting, health care professionals need to harness any opportunity to decrease time spent on unnecessary tasks such as medical image processing. Since there is a limited time period in which doctors can attempt to assist patients brought into casualty with newly sustained nerve damage, a decrease in medical image processing will be extremely beneficial to the patient and the medical staff. Since GPU implementation will shorten the time required to perform computationally intensive calculations,^52–54 cropping the data sets to the region of interest will further shorten the time required for processing (Fig. 3).²³ From the schematic, it is evident that despite the extra steps involved, the implementation of GPU systems will beneficially reduce the time required to process the data and diagnose the condition before patients undergo nerve replacement surgeries. In addition, 3D printing of the patient-specific C-N-S has the potential to provide the successful nerve damage therapeutic scaffold due to the fact that the C-N-S will be derived from high-resolution patient images.

FIG. 3.

Schematic displaying the proposed workflow that should be implemented in health care settings to diagnose potential nerve damage. This proposal provides a time-efficient solution to preparing patients for surgery, which can potentially aid in patients regaining the functioning of the nerve using the fabricated C-N-S.

Despite the foreseeable benefits of C-N-S implementation, there are several considerations that have to be highlighted. Since this proposal is for the usage of common medical imaging equipment, 3D medical scanners should be considered they are able to acquire free-form information from complex and hard-to-scan anatomical structures without the need to hold a defined pose. In addition, 3D scanners are also compatible with 3D printers and therefore need to be considered for nerve damage applications.^70,71 Due to the fact that that GPUs will accelerate 3D printer file slicing, it is possible that it may aid in lowering material loss and assist in overall 3D printing optimization. Since extrusion-based 3D bioprinters are known to have low resolution when compared to other printing methods,³⁵ technological advancements in 3D extrusion bioprinters are required to produce constructs with resolution at a similar quality to that of the medical images. In addition to this, widespread implementation of powerful GPU systems in desktop computers controlling the software for the chosen 3D bioprinter is required to implement this proposal. Although these may be costly considerations, the patients receiving the 3D bioprinted C-N-S will benefit from it with the possibility of improving their quality of life and well-being. Due to the inherent limited memory of GPU systems,²⁵ we propose the consideration and usage of computer-based servers, which could assist in supplying additional processing power, to beneficially decrease the time required for computation even more that GPU systems alone. Finally, although not covered in this technical note, we envisage the development of algorithms to provide a means of data checking to ensure that the printed structure is as close to “perfect” as possible. To achieve this, algorithms need to be developed to check the data file before the data are sliced in a 3D printing program. Subsequently, once the STL file has been generated, a second algorithm should be developed to verify whether each layer of the patient C-N-S corresponds in dimensions to the layers above and below it. If a discrepancy is found in either of the checks, the algorithm should send out a notification and automatically re-perform the problematic step as well as any step that follows it. In this way, despite requiring more seconds to re-perform the mistake, the time required for each step is minute and the data will still be processed in real time. Therefore, even with the mistake, the time required for image processing will still be rescued, while proving a more accurate scaffold to be printed.

The limitations to this proposal are that the proposed method is largely for hollow tube structures (tubular structures), whereas nerves are a solid cylindrical structure.⁵⁹ Despite this, the two-dimensional (2D) geometry as well as the volume of the tubular structure cylinder and the nerve cylinder will be similar. Due to the fact that the 2D structure is used to fabricate the 3D structure, the similarities in the 2D structure will allow for the translation from tubular image data processing to nerve damage image processing. Despite this, this limitation has to be kept in mind during data image processing. We propose the merger of algorithms for solid organ structures and tubular structures to overcome the hollow nature of lumen-based anatomical structures. Finally, the current regulatory challenges facing customized 3D printed constructs cannot be ignored.³⁰ However, this challenge should not limit the progression toward implementing this proposal.

Conclusion

The proposal of the merging of medical image cropping and 3D printing techniques provides a novel approach for providing patient-specific C-N-S for patients suffering with newly acquired nerve damage. Since nerve damage is particularly patient specific and nerve damage treatments remain unsuccessful, this proposal provides a possible opportunity to make a difference in patient care. Since medical imaging is the first tear in any nerve damage case, medical imaging processing time remains to be the limiting factor when it comes to patient care. The amalgamation of these concepts will see a beneficial decrease in the processing time of medical images, while providing patients with C-N-S for their individual injury specifications. Despite the foreseeable benefits, shortfalls in the current infrastructure will affect the implementation of this proposal. Despite this, we encourage researchers to attempt and implement the proposed structure to hopefully provide the much-needed solution to nerve damage.

Footnotes

Acknowledgement

This work was funded by the National Research Foundation (NRF) of South Africa.

Authors' Contributions

K.D.S. performed the analysis and collected the lab data; P.K., Y.E.C., and L.C.D.T. provided scientific and technical assessment as well as editorial input; and V.P. sourced the funding and conceptualized the study.

Disclosure Statement

No competing financial interests exist.

References

Steward

M.M.

, Sridhar

, and Meyer

J.S.

Neural regeneration. In: Heber-Katz

, Stocum

D.L.

, eds. New Perspectives in Regeneration. Current Topics in Microbiology and Immunology, Vol. 367. Berlin, Heidelberg: Springer, 2013, pp. 163–191.

Huebner

E.A.

, and Strittmatter

S.M.

Axon regeneration in the peripheral and central nervous systems. Results Probl Cell Differ, 48, 305, 2009.

Houschyar

K.S.

, Momeni

, Pyles

M.N.

, et al. The role of current techniques and concepts in peripheral nerve repair. Plast Surg Int, 2016, 1, 2016.

Richardson

P.M.

, McGuinness

U.M.

, and Aguayo

A.J.

Axons from CNS neurones regenerate into PNS grafts. Nature, 284, 264, 1980.

Benfey

, and Aguayo

A.J.

Extensive elongation of axons from rat brain into peripheral nerve grafts. Nature, 296, 150, 1982.

David

, and Aguayo

A.J.

Axonal elongation into peripheral nervous system “bridges” after central nervous system injury in adult rats. Science, 214, 931, 1981.

Guo

, Su

, Zeng

, et al. Reknitting the injured spinal cord by self-assembling peptide nanofiber scaffold. Nanomedicine, 3, 311, 2007.

Fenrich

, and Gordon

Canadian association of neuroscience review: axonal regeneration in the peripheral and central nervous systems—current issues and advances. Can J Neurol Sci, 31, 142, 2004.

Huang

Y.-C.

, and Huang

Y.-Y.

Biomaterials and strategies for nerve regeneration. Artif Organs, 30, 514, 2006.

10.

Pinho

A.C.

, Fonseca

A.C.

, Serra

A.C.

, Santos

J.D.

, and Coelho

J.F.J.

Peripheral nerve regeneration: current status and new strategies using polymeric materials. Adv Healthc Mater, 5, 2732, 2016.

11.

Williams

Central nervous system regeneration—where are we? QJM, 107, 335, 2014.

12.

Gordon

Nerve regeneration in the peripheral and central nervous systems. J Physiol, 594, 3517, 2016.

13.

Moattari

, Moattari

, Kaka

, Kouchesfahani

H.M.

, Sadraie

S.H.

, and Naghdi

Comparison of neuroregeneration in central nervous system and peripheral nervous system. Otorhinolaryngol Neck Surg, 3, 1, 2018.

14.

Shah

L.M.

, and Ross

J.S.

Imaging of spine trauma. Neurosurgery, 79, 626, 2016.

15.

Stoll

, Wilder-Smith

, and Bendszus

Imaging of the peripheral nervous system. In: Said

, Krarup

, eds. Handbook of Clinical Neurology, Vol. 115. Amsterdam: Elsevier B.V., 2013, pp. 137–153.

16.

Tins

B.J.

Imaging investigations in spine trauma: the value of commonly used imaging modalities and emerging imaging modalities. J Clin Orthop Trauma, 8, 107, 2017.

17.

Thakkar

R.S.

, Malloy

J.P.

, Thakkar

S.C.

, Carrino

J.A.

, and Khanna

A.J.

Imaging the postoperative spine. Radiol Clin North Am, 50, 731, 2012.

18.

Munera

, Rivas

L.A.

, Nunez

D.B.

, and Quencer

R.M.

Imaging evaluation of adult spinal injuries: emphasis on multidetector CT in cervical spine trauma. Radiology, 263, 645, 2012.

19.

Guarnieri

, Izzo

, and Muto

The role of emergency radiology in spinal trauma. Br J Radiol, 89, 20150833, 2016.

20.

Ohana

, Moser

, Moussaouï

, et al. Current and future imaging of the peripheral nervous system. Diagn Interv Imaging, 95, 17, 2014.

21.

Saito

, Ozawa

, Fujioka

, et al. Visualization of nerve fibers around the carotid bifurcation with use of a 9.4 Tesla microscopic magnetic resonance diffusion tensor imaging with tractography. Head Neck, 40, 2228, 2018.

22.

Rengier

, Mehndiratta

, von Tengg-Kobligk

, et al. 3D printing based on imaging data: review of medical applications. Int J Comput Assist Radiol Surg, 5, 335, 2010.

23.

Smistad

, Falch

T.L.

, Bozorgi

, Elster

A.C.

, and Lindseth

Medical image segmentation on GPUs—a comprehensive review. Med Image Anal, 20, 1, 2015.

24.

Kaur

, and Kaur

International journal of computer science and mobile computing various image segmentation techniques: a review. Int J Comput Sci Mob Comput, 3, 809, 2014.

25.

Smistad

, Elster

A.C.

, and Lindseth

GPU accelerated segmentation and centerline extraction of tubular structures from medical images. Int J Comput Assist Radiol Surg, 9, 561, 2014.

26.

Petcu

E.B.

, Midha

, McColl

, Popa-Wagner

, Chirila

T.V.

, and Dalton

P.D.

3D printing strategies for peripheral nerve regeneration. Biofabrication, 10, 032001, 2018.

27.

Giannitelli

S.M.

, Accoto

, Trombetta

, and Rainer

Current trends in the design of scaffolds for computer-aided tissue engineering. Acta Biomater, 10, 580, 2014.

28.

Seol

Y.-J.

, Kang

T.-Y.

, and Cho

D.-W.

Solid freeform fabrication technology applied to tissue engineering with various biomaterials. Soft Matter, 8, 1730, 2012.

29.

Ligon

S.C.

, Liska

, Stampfl

, Gurr

, and Mülhaupt

Polymers for 3D printing and customized additive manufacturing. Chem Rev, 117, 10212, 2017.

30.

Hourd

, Medcalf

, Segal

, and Williams

D.J.

A 3D bioprinting exemplar of the consequences of the regulatory requirements on customized processes. Regen Med, 10, 863, 2015.

31.

Donderwinkel

, van Hest

J.C.M.

, and Cameron

N.R.

Bio-inks for 3D bioprinting: recent advances and future prospects. Polym Chem, 8, 4451, 2017.

32.

Derakhshanfar

, Mbeleck

, Xu

, Zhang

, Zhong

, and Xing

3D bioprinting for biomedical devices and tissue engineering: a review of recent trends and advances. Bioact Mater, 3, 144, 2018.

33.

Kačarević, Ž., Rider

, Alkildani

, et al. An introduction to 3D bioprinting: possibilities, challenges and future aspects. Materials (Basel), 11, 2199, 2018.

34.

Zhang

, Gao

, Gu

, Yang

, Luo

, and Ma

High-resolution 3D bioprinting system for fabricating cell-laden hydrogel scaffolds with high cellular activities. Procedia CIRP, 65, 219, 2017.

35.

Ozbolat

I.T.

, and Hospodiuk

Biomaterials current advances and future perspectives in extrusion-based bioprinting. Biomaterials, 76, 321, 2016.

36.

Guvendiren

, Molde

, Soares

R.M.D.

, and Kohn

Designing biomaterials for 3D printing. ACS Biomater Sci Eng, 2, 1679, 2016.

37.

Aljohani

, Ullah

M.W.

, Zhang

, and Yang

Bioprinting and its applications in tissue engineering and regenerative medicine. Int J Biol Macromol, 107, 261, 2018.

38.

Jessop

Z.M.

, Al-Sabah

, Gardiner

M.D.

, Combellack

, Hawkins

, and Whitaker

I.S.

3D bioprinting for reconstructive surgery: principles, applications and challenges. J Plast Reconstr Aesthetic Surg, 70, 1155, 2017.

39.

Mandrycky

, Wang

, Kim

, and Kim

D.-H.

3D bioprinting for engineering complex tissues. Biotechnol Adv, 34, 422, 2016.

40.

Bishop

E.S.

, Mostafa

, Pakvasa

, et al. 3-D bioprinting technologies in tissue engineering and regenerative medicine: current and future trends. Genes Dis, 4, 185, 2017.

41.

Murphy

S.V.

, and Atala

3D bioprinting of tissues and organs. Nat Biotechnol, 32, 773, 2014.

42.

Yue

, Liu

, Coates

P.T.

, and Wallace

G.G.

Advances in printing biomaterials and living cells. Curr Opin Organ Transplant, 21, 467, 2016.

43.

Wang

, Abdulla

, Parker

, Samanipour

, Ghosh

, and Kim

A simple and high-resolution stereolithography-based 3D bioprinting system using visible light crosslinkable bioinks. Biofabrication, 7, 045009, 2015.

44.

Hsieh

F.-Y.

, and Hsu

3D bioprinting: a new insight into the therapeutic strategy of neural tissue regeneration. Organogenesis, 11, 153, 2015.

45.

Lee

S.-J.

, Esworthy

, Stake

, et al. Advances in 3D bioprinting for neural tissue engineering. Adv Biosyst, 2, 1700213, 2018.

46.

Knowlton

, Anand

, Shah

, and Tasoglu

Bioprinting for neural tissue engineering. Trends Neurosci, 41, 31, 2018.

47.

Mobini

, Spearman

B.S.

, Lacko

C.S.

, and Schmidt

C.E.

Recent advances in strategies for peripheral nerve tissue engineering. Curr Opin Biomed Eng, 4, 134, 2017.

48.

Oudega

, Moon

L.D.F.

, and de Almeida Leme

R.J.

Schwann cells for spinal cord repair. Brazilian J Med Biol Res, 38, 825, 2005.

49.

Mayo Clinic Staff. Spinal cord injury. Available at https://www.mayoclinic.org/diseases-conditions/spinal-cord-injury/diagnosis-treatment/drc-20377895 (Last accessed November 9, 2018).

50.

Clinic

Paralysis: diagnosis and tests. Available at https://my.clevelandclinic.org/health/diseases/15345-paralysis/diagnosis-and-tests (Last accessed November 9, 2018).

51.

Chin

L.S.

, Mesfin

F.B.

, and Dawodu

S.T.

Spinal cord injuries. Available at https://emedicine.medscape.com/article/793582-overview#a1 (Last accessed November 9, 2018).

52.

Pratx

, and Xing

GPU computing in medical physics: a review. Med Phys, 38, 2685, 2011.

53.

Castaño-Díez

, Moser

, Schoenegger

, Pruggnaller

, and Frangakis

A.S.

Performance evaluation of image processing algorithms on the GPU. J Struct Biol, 164, 153, 2008.

54.

Eklund

, Dufort

, Forsberg

, and LaConte

S.M.

Medical image processing on the GPU—past, present and future. Med Image Anal, 17, 1073, 2013.

55.

Gibson

, Giganti

, Hu

, et al. Automatic multi-organ segmentation on abdominal CT with dense V-networks. IEEE Trans Med Imaging, 37, 1822, 2018.

56.

Dalca

, Danagoulian

, Kikinis

, Schmidt

, and Golland

Segmentation of nerve bundles and ganglia in spine MRI using particle filters. Med Image Comput Comput Assist Interv, 14, 537, 2011.

57.

Levin

H.S.

, Amparo

, Eisenberg

H.M.

, et al. Magnetic resonance imaging and computerized tomography in relation to the neurobehavioral sequelae of mild and moderate head injuries. J Neurosurg, 66, 706, 1987.

58.

Levine

P.A.

, Paling

M.R.

, Back

W.C.

, and Cantrell

R.W.

MRI vs. high-resolution CT scanning: evaluation of the anterior skull base. Otolaryngol Neck Surg, 96, 260, 1987.

59.

Sunderland

S.S.

The anatomy and physiology of nerve injury. Muscle Nerve, 13, 771, 1990.

60.

Bücking

T.M.

, Hill

E.R.

, Robertson

J.L.

, Maneas

, Plumb

A.A.

, and Nikitichev

D.I.

From medical imaging data to 3D printed anatomical models. PLoS One, 12, e0178540, 2017.

61.

Reynisson

P.J.

, Scali

, Smistad

, et al. Airway segmentation and centerline extraction from thoracic CT—comparison of a new method to state of the art commercialized methods. PLoS One, 10, e0144282, 2015.

62.

Lesage

, Angelini

E.D.

, Bloch

, and Funka-Lea

A review of 3D vessel lumen segmentation techniques: models, features and extraction schemes. Med Image Anal, 13, 819, 2009.

63.

, van Ginneken

, Reinhardt

J.M.

, et al. Extraction of airways from CT (EXACT’09). IEEE Trans Med Imaging, 31, 2093, 2012.

64.

Narayanaswamy

, Dwarakapuram

, Bjornsson

C.S.

, Cutler

B.M.

, Shain

, and Roysam

Robust adaptive 3-D segmentation of vessel laminae from fluorescence confocal microscope images and parallel GPU implementation. IEEE Trans Med Imaging, 29, 583, 2010.

65.

Yang

, Yang

J.D.

, Hwang

H.P.

, et al. Segmentation of liver and vessels from CT images and classification of liver segments for preoperative liver surgical planning in living donor liver transplantation. Comput Methods Programs Biomed, 158, 41, 2018.

66.

Yang

, Kitslaar

, Frenay

, et al. Automatic centerline extraction of coronary arteries in coronary computed tomographic angiography. Int J Cardiovasc Imaging, 28, 921, 2012.

67.

Okada

, Linguraru

M.G.

, Hori

, et al. Multi-organ segmentation in abdominal CT images. In: 2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Vol. 100. IEEE, 2012, pp. 3986–3989.

68.

Smistad

E. G

, PU-Based Airway Tree Segmentation and Centerline

Extraction

, 2012. Available at https://hgpu.org/?p=8517 (Last accessed July 8, 2019).

69.

Liu

, Telea

A.C.

, Roerdink

J.B.T

.M., et al. Parallel centerline extraction on the GPU. Comput Graph, 41, 72, 2014.

70.

Straub

, and Kerlin

Development of a large, low-cost, instant 3D scanner. Technologies, 2, 76, 2014.

71.

Haleem

, and Javaid

3D scanning applications in medical field: a literature-based review. Clin Epidemiol Glob Heal, 2018; DOI:10.1016/j.cegh.2018.05.006.