Study design and patients
Patients with tibial fracture treated by TSF in Tianjin Hospital were retrospectively analyzed from January 2016 to June 2019. The inclusion criteria were: (1) comminuted fracture (AO/Asif classification C3); (2) compound fractures (Gustilo type II / III); (3) the follow-up time after frame removal was ≥6 months. Exclusion Criteria were: (1) Patients with bilateral tibia fractures (unable to provide the mirror image of the contralateral three-dimensional reconstruction image); (2) patients unable to cooperate with regular follow-up. Finally, 41 patients were included in the study. There were 21 patients in the marker-3D measurement group (experimental group) and 20 patients in the traditional radiographic measurement group (control group).
Measurement methods
All the treatment procedures were performed by the same surgical team. All patients underwent TSF installation as follows: the frame was fixed to the bone segment firstly with the struts in a sliding state, the fracture was preliminarily reduced by moving the rings under the C-arm followed by the lock of struts, residual deformities were corrected postoperatively by adjusting these struts.
Traditional radiographic measurement
The postoperatively standard radiographs (AP and lateral radiographs, including proximal and distal joints as much as possible) of patients were conducted. The X-rays were imported into computer for parameters measurement (Fig. 1). The proximal bone segment was used as the reference, and the distal bone segment was determined as the free movement end. The midpoint of the proximal fracture line was taken as the center of rotation of angulation (CORA).
Parameters need to be measured include six deformity parameters and four mounting parameters according to the instructions. The deformity parameters include angulation and translation in the coronal (α1, S1), sagittal (α2, S2), and axial plane (physical examination, T). The mounting parameters which describe where the center of the reference ring is located relative to the origin point include anteroposterior view frame offset (L1), lateral view frame offset (L2), axial view frame offset (L3), and the rotary frame angle (physical examination) (Fig. 1).
Marker-3D measurement method
The marker was a composite structure, which was composed of aluminum alloy marker ball and photosensitive resin connecting rod (Fig. 2). A set of same markers were used for measurement.
3D reconstruction
Three markers were mounted on the proximal ring, the other three were mounted on the distal ring. The markers were distributed on each ring as evenly as possible (120 degrees). Bilateral lower limbs of each patient in the marker-3D measurement group underwent CT (GE Optima, CT66) scan for 3D reconstruction. The following models were generated: the 3D model of the proximal bony fragment of the affected limb (Model Proximal), the 3D model of the distal bony fragment of the affected limb (Model Distal), the 3D mirror model of healthy limb bone (Model Reference), the 3D model of external fixator (Model Frame), and the 3D model of Marker Balls (Model Marker Balls) (Fig. 3). The 3D mirror model of healthy limb was used for registration [21, 22].
Preparation in software
The proximal/distal bony fragment and its relative ring was considered a rigid part respectively. A self-designed 3D reduction software was used for the measurement of electronic prescription (Fig. 4). The detailed marker locations on the ring needed to be inputted into the reduction software in which the spatial position of the marker balls can be recognized automatically as well as the initial position of the two rings. During the virtual fracture reduction, the software could automatically record the change in position.
Virtual fracture reduction using the custom software
(1) The reconstructed 3D models and the information of the frame and markers were imported into the custom software for virtual fracture reduction. (2) The protuberance tip on the bony segment and the feature point on the joint were used as the references. The Model Proximal was considered as the fixed end, the Model Distal was registered with the Model Reference to achieve fracture reduction directly. It was also possible to add multiple reduction intermediate points according to requirements. “traction-rotation-alignment” was the motion path of the bone to ensure the reduction safety. (3) The software could automatically generate the reduction path of the free movement end avoiding the collision between the bony segments. Furthermore, the virtual reduction animation was generated according to the initial and final position of the fixed ring. (4) The relative position changes of the two rings could be determined according to step 3. The length changes of the six struts were calculated by the Stewart mechanism kinematics algorithm, the strut ‘s adjustment plan (electronic prescription) was obtained finally. The schematic diagram was shown in Fig. 5. Typical case was shown in Fig. 6.
Effectiveness evaluation
The effectiveness was evaluated by the residual displacement deformity (RDD) and residual angle deformity (RAD) in the coronal and sagittal plane, according to the standard AP and lateral X-rays after reduction.
Statistical analysis
SPSS 22 (IBM Inc., New York, USA) was used for statistical analysis. The comparison between age was conducted by Student’s t test and represented as −x ± s. The categorical data was compares by Chi-square test. The measurement data of abnormal distribution (residual deformities) was expressed as M (P25, P75) followed by Mann-Whitney U test. Significant difference was set as P ≤ 0.05.