The creation of the patient-specific 3D model of a meniscus has eight main steps starting from the medical imaging of patient’s knee which are overviewed herein.

Step 1. The MRI acquisition: Static medical imaging of the patient’s knee is the starting point of the patient-specific meniscal scaffold development [10], but dynamic MRI can be a future approach [24]. MRI is the most used medical imaging method for the knee in the clinics and also safer than the CT for the patient. As diagnosed by the patient’s orthopedic surgeon, the implanted-needing leg of the patient is scanned in feet-first supine position preferably with sagittal plane as the acquisition plane either as T1- or T2-weighted MRI. The routine MRI acquisition is not sufficient to extract the meniscus; therefore, a specific imaging protocol should be used. The acquisition should have a 3D isotropic sequence providing a high-spatial-resolution Digital Imaging and Communications in Medicine (DICOM) dataset. Isotropic sequence means that the volumetric images have the same resolution in all dimensions. This MRI acquisition will take longer than the routine ones. This brings two challenges in the MRI room: increased cost and keeping the patient still during the entire scan since each time when the patient is not still, the acquisition must be repeated. The technical details should be well communicated between the MRI technician and the tissue engineer who will perform the meniscus extraction and manufacture the scaffold. The challenge in this step is that the extracted 3D model is directly coming from the patient’s MRI. Thus, if the patient’s MRI is from a date that the patient needs a meniscus implant, probably the obtained 3D model of the meniscus will not be the model of the healthy meniscus. If there are any changes in the knee osteochondral tissues, probably the size and shape of the original healthy meniscus will be lost. Since meniscus surgeries are the most frequent orthopedic procedures, perhaps when the ideal meniscus implant is developed by the tissue engineers, a dramatic change will happen in the health-care system and/or in the health insurance system.

Step 2. Digital extraction of the meniscus: Digital extraction of the meniscus through medical image segmentation is a critical step in the manufacture of a patient-specific meniscal scaffold [10]. Digital extraction means the segmentation of the meniscus tissue in the knee DICOM images that came from the MRI. There are several tools for segmentation including but not limited to MITK [25], GIST [25], and Analyze [26]. Cengiz et al. [10] have reported in detail the use of an interactive, real-time geodesic level-set algorithm, a variation of the level-set algorithm to extract meniscus for the use of meniscus tissue engineering (Fig. 32.2a–c). Besides, when using a semiautomatic segmentation method, manual corrections on the segmented images are needed. This includes some manual refinements to increase the accuracy of the meniscus extraction. Depending on the obtained DICOM dataset, an image preprocessing might be needed, for example, cleaning of the undesired artifacts from the images because of acquisition irregularities. This can be the removal of low- or high-intensity values to keep a particular grayscale range. This preprocessing can be performed by custom-written codes using the Insight Segmentation and Registration Toolkit (ITK) open-source libraries [27].

Step 3. 3D meniscus model creation: The patient-specific 3D meniscus model reconstruction [10] can be obtained by the segmented images using surface rendering and more specifically by using the marching cubes algorithm [28]. This can be performed with a software once the digital extraction of the meniscus is achieved from the DICOM dataset. Figure 32.2d shows an obtained patient-specific 3D meniscus model.

Step 4. 3D model post-processing: 3D tissue models extracted from medical images are in fact not to be further processed to remain as the original tissue in terms of shape and size. However, while being still patient specific, the 3D meniscus model can be smoothened and minor holes can be filled to avoid practical issues regarding scaffold production as a post-processing step. Based on the tool used, the extraction of the meniscus from the DICOM dataset can provide a 3D model in the form of a triangle mesh in Visualization Toolkit (VTK) file format [10]. The 3D model should be converted into stereolithography (STL) format that is the standard format for most 3D printing.

Step 5. Biomaterial development: The complete process of development of biomaterial for a meniscus scaffold is a tremendous amount of work that takes many years, so the selection of biomaterial on the overall implant development process is of great importance. Most of the natural biomaterials have a low potential for direct 3D printing, while synthetic biomaterials have much higher potential for direct printing. Natural biomaterials [29, 30] could be however processed via indirect 3D printing while still being patient specific. This is the reason why biomaterial development step (shown in green) is located prior the post-processing of the 3D meniscus model. In the case that the scaffold is to be produced with an indirect 3D printing, then there are two main options:

Scaffold development process involves the characterization of in vivo performance of the scaffold using animal models [31]. Thus, it is also important for animals to receive meniscal scaffolds that are specific for that particular animal (Fig. 32.4) for the scaffold to maximize its performance.

Step 6. Scaffold manufacturing—alignment and layering: This step is about the 3D virtual alignment of the meniscus model or the mold model. In the case where scaled-up production is needed, multiple objects are inserted and aligned to be manufactured simultaneously. This brings an important advantage by shorting the fabrication time per object by avoiding the waiting time between layers to allow solidification. The layering of an object defines the total number of layers to be fabricated. However, in fact, the used bio/material’s printability combined with the specified printing parameters provides the layer thickness [10]. Thus, the model should be layered accordingly as depicted in Fig. 32.5.