A 30-year-old healthy male with a body height of 172 cm and a body weight of 75 kg was selected as a volunteer. The subject had no history of femoral fracture or trauma, and femoral lesions and injuries were also ruled out by X-ray imaging. The spiral axis scan of the bilateral femurs was performed by using a CT scanner (Optima CT 660; GE Healthcare) from top to bottom using the scanning condition of 120 KV, 125 mA and a layer thickness of 1 mm. The above machine had an advanced image post-processing and camera function.

After the original data were interpolated and enlarged on the machine, a continuous picture of the layer thickness was obtained, and was burned to the optical disc in the international standard dicom format. The study was approved by the Ethics Committee of The Third Hospital of Hebei Medical University (Shijiazhuang, China). The internal fixation model was built using Solidwork 2016 software (Dassault); the novel intramedullary nail system was designed by the Third Hospital of Hebei Medical University (patent number, ZL201721394941.3).

The femur computed tomography (CT) image (dicom format) was imported into the Mimics 17.0 software (Materialise) to segment bonefrom the entire model bythresholding; the images were subsequently selectively segmented using ‘split-mask’ according to the anatomical structure. Finally, a geometric model of the femur was established through regional expansion, smoothing all parts of the femur, filling small gaps, and making the outer contour of the femur smooth and continuous.

The constructed femur information was saved in an STL format file, and the STL format file was repaired and optimized by reverse engineering software Geomagic Studio 2012 (Raindrop). In brief, the non-characteristic masses and indentations on the surface of the model were removed and the surface was smoothed. Furthermore, non-characteristic high curvature and the generation of self-intersecting faces were prevented. The smooth surface of the model was used to fit the triangular surface of the model surface; the femoral cortical bone and cancellous bone were distinguished by using the offset function. At last, a continuous curved surface model was generated.

It was assumed that the fracture surface was completely broken and in a contact state. The friction coefficient was 0.2 according to a previous study (7). The femur was subjected to numerous loads, including muscle force, joint force and dynamic impact force. In view of numerous uncertainties, including the number and direction of load during finite element simulation of muscle loading, it was relatively complex to accurately simulate the loading conditions in the dynamic state (8). The simple model method of Duda et al (9) was used in the present study. To simulate the load of the femur while standing and walking, analysis ofthe static mechanics of the model was performed. The condition was that the distal knee joint had a degree of freedom of 0 and the distal nodes have a displacement of 0 on the X, Y and Z axes. Furthermore, a load of 700 N along the direction of the force line at the outer surface of the contact area between the femoral head and the acetabulum was applied.

According to Duda et al (9), the intact femur received a torque of ~0.04 body weight meters (BWM) during the gait cycle. However, three weeks after the intramedullary nailing of the femoral shaft fracture, the maximum torque of the femur was 0.01 BWM (10). Therefore, based on the load of 700 N, the maximum torque of the femur was 7-28 Nm. Therefore, in the torsion analysis under exclusion of the specific effects of each muscle group, the boundary load condition was fixation on the femoral condyles and a 7-Nm external rotation torque was applied at the proximal end of the femur along the same axis as compressive loading.

In general, biological tissue belongs to an anisotropic non-linear body. The application of bone finite element analysis in the field of biomechanics is mostly based on the isotropic, homogeneous continuous linear elastomer. Therefore, the model utilized in the present study assumed that the femoral bone is an isotropic, homogeneous linear elastic material, which consists of cortical bone, cancellous bone and internal fixation (Table I), and also referred to the data reported by previous studies (11,12). Only two materials of bone and internal fixation were considered in the model, and muscle and tendon stress were simplified, regardless of the friction between the joints and cartilage. A femoral geometric model was established by using Mimics 17.0 (Materialise) and Geomagic Studio 2012 (Raindrop) software based on the CT image of the femur (Fig. 1).

The traditional intramedullary nail system (Fig. 2) was based on the intramedullary nail (diameter, 10 mm; length, 380 mm) produced by Double Medical Technology Co., Ltd. (Fig. 2Aa). Solidwork 2016 (Dassault) is a software for simulation design and has been applied in numerous fields, including aviation and machinery, where the modeling results were reliable (6). The internal fixation model was built using the Solidwork 2016 (Dassault) software, a novelintramedullary nail system designed in The Third Hospital of Hebei Medical University in 2017 (patent no: ZL201721394941.3; Fig. 2Bb). The geometric structure and the specific parameters of the novel intramedullary nailing system are provided in Fig. S1. It is designed as a hollow nail body, including the head, body and tail of the nail. The nail body includes a slotted section and a corresponding connecting section. The upper ring of the side wall of each grooved section has several grooves connected to the mouth of the nail body, which are spirally arranged along the extension direction of the nail body.

The femoral solid model in IGES format generated by Geomagic Studio 2012 (Raindrop) was imported into the design software Solidwork 2016 (Dassault) to simulate femoral shaft fractures (Fig. 2Ac and Bc). Combined with the above internal fixation model, two different internal fixation femoral shaft fractures in three-dimensional finite element models were established (Fig. 2Ad and Be).

After verifying the validity of the above traditional and novel femoral finite element models (Fig. 2A and B), the femoral comminuted fractures were fixed with the traditional and novel intramedullary nail systems. Subsequently, the finite element method was used to analyze a complete femoral model under a 700-N load. The stress cloud map (Fig. 3A) and the displacement cloud map (Fig. 3B) were also provided. The stress was mainly distributed in the middle and lower part of the femur, around the femoral neck and the large and small trochanter. There was also an obvious stress concentration at the small trochanter of the femur (Figs. 4-6), prior to healing as well as after healing.

Under the compressive load, the maximum stresses, the maximum displacements, broken bone stresses and broken bone displacements among the traditional and the novel intramedullary nail fixation prior to healing, the traditional and novel intramedullary nailing after healing, and the complete femur were also provided in Table II. Under torsional load, the traditional and novel intramedullary nail fixations prior to healing were also compared (Table III).

When subjected to compressive or torsional load prior to complete healing, the stability of the novel IM nail fixation was greater compared with that of the traditional IM nail fixation, as indicated by the smaller resulting maximum fracture site displacement (Table III). The novel IM nail shares more loads as indicated by the higher maximum stress in the novel intramedullary nail group compared with the traditional IM nail group (Table III). The fragment stress in the novel IM nail under both loading conditions were higher because of the higher load concentration around Kirschner wires in the fragment.

It was suggested that the novel intramedullary nail fixation system was superior compared with traditional intramedullary nail fixation systems with regards to maximum stress and displacement, which stimulates bone growth.