Skin is composed of a fibrous network rich in collagen, embedded within a ground substance matrix. Its mechanical behavior is nonlinear viscoelastic, exhibiting anisotropic characteristics depending on the orientation of the fibrous network. The mechanical properties of skin vary among individuals, influenced by factors such as gender and age, and change over time (Wijn et al., 1976; Lanir, 1981; Smith et al., 1982; Stark, 1977). It is worth noting that, in addition to these factors, the mechanical properties of skin also depend on the methods used to obtain the data. Various experimental techniques, including indentation, torsion, suction, uniaxial and dynamic testing, and optical methods, have been used to characterize human skin. Yazdi and Baqersad (2022) reviewed the literature studies published from 1969 to 2021, analyzing 130 papers focused on the study of mechanical properties of human skin. They summarized and compared the skin data obtained through different experimental methods. The reported mechanical properties varied significantly among studies. For example, the Young’s modulus of forearm skin can vary by more than 50,000-fold between its lowest and highest reported values. Using an in vivo indentation test, Bader and Bowker (1983) measured a Young’s modulus of 1.09 kPa in the forearm skin of a young female subject. In contrast, Grahame and Holt (1969) reported values ranging from 18 to 57 MPa using an in vivo suction experiment, highlighting how skin stiffness changes with aging. The Young’s modulus also varies depending on the location of the skin on the human body. For instance, a modulus of 83.3 MPa was reported for back skin (Yazdi and Baqersad, 2022; Ní Annaidh et al., 2012) through uniaxial tensile ex vivo tests. This value was significantly different from that measured in forearm skin by Bader and Bowker (1983). Additionally, it was also found that compared to normal skin, the scar tissue has increased stiffness (Dunn et al., 1985; Corr and Hart, 2013).

The large variances in Young’s modulus are attributed to experimental errors, the high heterogeneity of human skin, and its nonlinear mechanical behavior. This wide variability and uncertainty present a challenge for FEA in accurately predicting skin behavior for patient-specific skin flap design as documented values may differ significantly from those of an individual patient. To model skin mechanics, researchers have applied various approaches. Table 1 compares different skin modeling approaches used in FEA for flap design. Some studies simplified skin as an isotropic, linear elastic material, while others represented it as a nonlinear, isotropic, hyperelastic material, deriving parameters through experimental data fitting. Additionally, certain studies accounted for the orientation of collagen fibers, incorporating anisotropic properties. Other tissues (e.g., bone), although mentioned in a few papers such as Retel et al. (2001), were not included in this section as they are not directly relevant to the application of FEA in skin flap design.

As stated previously, the complex microstructure of skin, coupled with significant variability in its mechanical properties across individuals and different anatomical locations, poses challenges for the numerical analysis of skin flaps. Furthermore, measurement techniques and subject-specific factors contribute to the difficulty of achieving accurate modeling. To navigate these complexities, researchers have introduced various assumptions and simplifications in skin flap material modeling, leading to diverse material models with distinct parameters as shown in Table 1. In essence, the skin flap properties used in FEA rely heavily on theoretical predictions found in the literature, each incorporating different aspects with unique assumptions. This variability in modeling approaches hinders the development of patient-specific flap designs, in contrast to the more predictable FEA outcomes observed in implant studies. Unlike engineering materials, which possess stable and well-defined properties that can be represented using straightforward material models, skin is a living tissue with highly dynamic mechanical responses. This makes personalized modeling particularly difficult.

Therefore, obtaining more accurate skin properties remains a key challenge for researchers and clinicians working in the field. As extracting a skin sample near a wound for mechanical testing to inform FEA might not be a practical option in clinical settings, the solution may lie in developing non-invasive techniques to accurately assess individual skin characteristics. Recent advancements also highlight the need for contact-free testing methods to evaluate the mechanical behavior of human skin. Various imaging modalities, such as optical coherence tomography (OCT), photoacoustic tomography (PAT), and reflectance confocal microscopy (RCM), have been used for skin analysis, offering promising approaches for non-invasive assessment (Lentsch et al., 2022).

Based on the findings, future research should focus on refining non-invasive techniques to obtain patient-specific mechanical property parameters before surgery, which can then serve as the input for FEA, as also reported by Myoung et al. (2021). With advancements in imaging technologies, such as high-speed cameras, obtaining skin parameters for surgical quantification is becoming increasingly feasible. For example, Myoung et al. (2021) introduced the Swing anglemeter, a device designed to assess skin elasticity. Their study involved 45 healthy Korean women aged 23–60 years, using a rubber ball dropped onto the subjects’ cheeks and tracking its rebound trajectory. A mobile phone camera was used to capture the maximum rebound angle for each test, which was then correlated with skin elasticity. Although this method only provides a general assessment of skin condition rather than direct mechanical parameters for numerical analysis, it suggests a promising path toward a non-invasive, user-friendly approach for property estimation before surgery, potentially enabling the quantification of flap surgery procedures.

Imaging technologies such as axial computed tomography (CT) scans and magnetic resonance imaging (MRI) have been widely used in the medical field since the last century for diagnosis, surgical planning, and post-procedural follow-ups (Vannier and Marsh, 1996). Computer-assisted surface imaging techniques have been used in plastic and reconstructive surgery since the late 20th century (Marsh et al., 1986). In the 21st century, 3D scanning methods have become increasingly prevalent for surface scanning, enabling the collection of detailed surface data for plastic reconstruction. For example, the commercially available FaceSCAN 3D (3D-Shape GmbH, Erlangen, Germany) was used to capture 3D images, which were then applied to evaluate the aesthetic outcomes of nasal plastic reconstruction (Peters et al., 2021). With advancements in computational power and numerical analysis software during the 2000s, numerical simulations became both feasible and practical for assisting in plastic surgery procedures, including syndactyly reconstruction. These imaging techniques have since been integrated into the construction of geometric models for FEA, enhancing the precision and effectiveness of surgical planning and evaluation.

In the preprocessing stage of FEA of skin flap, researchers must make several critical decisions, including geometry construction, meshing strategies, defining connections between different parts of the model, and selecting software used for geometry creation, meshing, and solving. We reviewed these aspects based on selected studies and summarized them in Table 2 for comparison.

Table 2 shows that some studies have analyzed the effect of skin flap parameters using predefined geometries in FEA. In these cases, no imaging conversion was required as the geometries could be easily generated using the software’s modeling functions. Miyamoto et al. (2010) constructed a 3D model based on CT scans; however, they did not simulate the suture closure process. Instead, they treated the skin flap and underlying soft tissue as a single solid object. Ji et al. (2024a) used a laser scanner to capture the wound surface data, obtaining the 3D geometry of the wound surface. They then applied an algorithm they had developed to unfold the 3D surface into a 2D representation (Ji et al., 2024b). Notably, their study explicitly stated the contact condition between the hand’s bone, skin tissue, and grafted skin flap as frictionless contact. This contrasts with other studies, where some researchers used a plane stress condition without considering the underlying structures.

Despite significant advancements in 3D imaging techniques, 3D surface scanning has not been widely applied in FEA for flap design. This may be due to the irregular shape of the human body surface, particularly in hand anatomy, making the numerical modeling of skin flap coverage and closure challenging. It is important to note that effectively integrating FEA into patient-specific flap design for complex surfaces and shapes requires iterative adjustments to achieve an optimal surgical plan. The numerical difficulty of such a 3D contact problem would make it challenging to obtain convergent results. Therefore, unfolding an irregular 3D surface into a 2D plane appears to be a promising approach, as demonstrated by Ji et al. (2024b). Given the complex contours of the human body—especially the hand—3D scanning, combined with computational techniques to flatten the surface, could greatly enhance feasibility. This method would facilitate flap geometry construction, improve mesh quality, reduce computation time, and simplify result visualization and analysis. When studying the stress and strain in a skin flap stretched over a wound, flattening the surface would also facilitate the application of boundary and loading conditions.

However, presenting the FEA results in 2D might not be intuitive to everyone, especially for surgeons, who are accustomed to working with 3D representations of anatomical structures during surgery. Future work could focus on developing methods to transition from 2D to 3D surface visualization. MR and AR technologies are well-suited for this task as they can provide real-time, interactive 3D visualizations that transform the 2D results into a more intuitive 3D representation, even restoring the original shape of the flattened surface. This would allow clinicians to view and interact with the flap design in its true anatomical form while still benefiting from the simplified 2D representation for analysis. There were proposed algorithms that deform or bend flat surfaces into 3D curved surfaces, which provide valuable insights into how 2D surfaces can be transformed into 3D shapes (Jin et al., 2022; Sam et al., 2024). Although these algorithms have not been specifically tested in MR and AR devices, these platforms possess the same computational capacity as regular computers, allowing for their implementation in real-time, interactive environments. This should be evaluated in future research.

The skin flap surgery involves wound closure with suture stitches using various flaps. This procedure can be simulated by applying appropriate boundary and loading conditions during the preprocessing stage of FEA. Once the preprocessing is completed, FEA can be conducted. The FEA results (i.e., postprocessing variables) include force, displacement, stress, strain, and other variables. Therefore, the specific results to be considered and generated in a study must be carefully selected based on the objectives and requirements of the analysis. Table 3 summarizes the simulated procedures, boundary and loading conditions, and postprocessing variables applied in the studies. It shows that some researchers accounted for the pretension of the skin surrounding the wound before surgery by applying prestress to simulate this effect, while others did not. The boundary and loading conditions generally involve either fixation or displacement control to simulate the suture process for wound closure. Another observation is that, although not explicitly stated in the papers, all the analyses appear to be static. This is consistent with expectations as no dynamic behavior is considered in the simulation of wound closure. Some studies validated their FEA results by comparing them with clinical observations, theoretical models, or surrogate data—for example, by identifying consistency with known stress patterns in skin necrosis or flap surgery. Others did not include formal validation but still provided valuable insights through trend analysis, such as examining the influence of varying material parameters. These trend-focused studies remain meaningful despite the lack of validation as their goal was comparative analysis rather than precise clinical prediction.

In summary, these studies primarily examined closure or suture forces and stresses within the skin flap. Although uncertainty in skin material properties impacts the reliability of the results, the findings still provide valuable insights for trend analysis, particularly when comparing the effects of geometric and material parameters.

There is a current trend of applying novel visualization approaches, such as MR and AR, to integrate FEA results into surgical workflows. Although other means, such as colormaps and software interfaces, exist, MR and AR offer enhanced real-time interaction and immersive visualization. This helps medical professionals better interpret engineering principles and results more effectively in surgical decision making.

A common use of MR and AR visualizations integrating FEA results in surgery is to provide real-time representations of organ or soft tissue deformation (Golse et al., 2021; Kong et al., 2017; Perruisseau-Carrier et al., 2017; Nikolaev and Cotin, 2020; Samei et al., 2018; Kugler et al., 2017). For example, AR systems have been developed and evaluated to assess the feasibility of dynamically updating liver deformation (Golse et al., 2021; Nikolaev and Cotin, 2020; Kugler et al., 2017). By integrating AR with FEA results, this approach allows the liver model in AR to align with the organ’s shape in real time during surgery. CT scans were usually obtained to reconstruct a 3D organ model preoperatively, which allows for precise anatomical segmentation and patient-specific surgical planning. By integrating an FEM, the reconstructed organ model can dynamically adapt to intraoperative deformations (Golse et al., 2021; Nikolaev and Cotin, 2020; Kugler et al., 2017). This ensures accurate alignment with real-time surgical conditions, thereby providing precise localization of tumors and resection planning (Nikolaev and Cotin, 2020), along with real-time surgical guidance (Kugler et al., 2017). A similar approach was used in a study that developed an AR-guided navigation system for precise intraoperative tumor localization during laparoscopic kidney surgery (Kong et al., 2017). The virtual organ model dynamically updates in real time to reflect deformations caused by surgical manipulation. Prostate deformation was also visualized in an AR application that integrated FEM with real-time transrectal ultrasound (TRUS) imaging to enhance intraoperative navigation during robot-assisted laparoscopic radical prostatectomy (RALRP) (Samei et al., 2018). This approach also improved tumor localization and surgical precision (Samei et al., 2018). In addition to organs, Perruisseau-Carrier et al. (2017) focused on applying patient-specific MRI data for the 3D reconstruction of nerve structures and then integrating FEA to simulate nerve deformation under physiological and surgical conditions. AR overlaid the updated FEM nerve model onto the surgical field, allowing surgeons to visualize real-time nerve deformations and adjust their approach accordingly.

There were also studies that have leveraged the capability of MR and AR to provide a sense of depth (Vörös et al., 2023; Tonutti et al., 2017). Vörös et al. (2023) investigated the use of a 3D autostereoscopic display in a simulated laparoscopic task to restore depth perception for surgeons. By integrating an MR simulator with an FEM, the system allowed participants to visualize soft tissue deformation in 3D and improve task performance compared to conventional 2D visualization. The results demonstrated significant reductions in task completion time, instrument travel distance, and error rates. This highlighted the benefits of depth-enhanced visualization in minimally invasive surgery. In addition, a machine learning approach was proposed to enhance visualization by allowing the surgeon to visualize internal structures with depth perception through an AR headset (Tonutti et al., 2017). By integrating pre-computed FEA results with artificial neural networks (ANNs) and support vector regression (SVR), the system accurately modeled soft tissue deformation under varying loads. This enabled real-time updates of tumor position with errors below 0.3 mm, demonstrating significant potential for assisting surgeons in more precise tumor localization and resection.

From the abovementioned descriptions, AR and MR can be regarded as wearable computing systems that enable real-time computation and optimization. Studies have also demonstrated their capability to integrate machine learning and deep learning algorithms to predict various surgical outcomes and anatomical changes dynamically. For example, a study applied AR to predict the final shape of a catheter during endovascular procedures by integrating fiber Bragg grating (FBG) sensors with FEA results (Scarponi et al., 2024). By leveraging machine learning algorithms, the system continuously refined predictions, enhancing surgical precision and reducing reliance on repeated fluoroscopic imaging. Machine learning algorithms have been used to accelerate FEA-based deformation predictions for real-time use (Samei et al., 2018), which is critical in intrasurgical applications.