Patients

The institutional review board of our hospital approved this study (Number 20150129). All patients provided written informed consent. From April 2014 to October 2019, patients with complete follow-up data who underwent internal fixation surgery for lower extremity fractures at our hospital were recruited. The follow-up data included X-ray and CT data 9months after surgery. Two senior orthopedic doctors and an imaging physician judged the degree of fracture healing in the patients according to the patients X-ray, CT and clinical data. According to the above criteria, the status was judged as bone healing, poor bone healing and bone nonunion. If the diagnosis was inconsistent among the three doctors, the same result provided by two of the doctors was taken, and if the results of the three doctors were not the same, the case was excluded. A total of 79 patients were included in the study. A total of 112 CT scans were performed, 49 of the scans were judged to show bone healing (group A), 37 were judged to show poor bone healing (group B), and 26 were judged to show nonunion (group C). There were 21 females in group A, with an average age of 46.712.4years and a follow-up time of 18.73.4months. There were 12 cases of femoral fracture, 37 cases of tibial fracture, 15 cases of intramedullary nail fixation, and 34 cases of plate screw fixation in group A. There were 15 females in group B, with an average age of 51.214.5years and an average follow-up time of 14.73.8months. There were 9 cases of femoral fracture, 28 cases of tibial fracture, 24 cases of plate fixation, and 13 cases of intramedullary nail fixation in group B. There were 15 females in group C, with an average age of 49.213.5years and an average follow-up time of 15.74.4months. There were 8 cases of femoral fracture, 18 cases of tibial fracture, 17 cases of plate fixation, and 9 cases of intramedullary nail fixation in group C (Table1). The fractures of the samples were diaphyseal and metaphyseal, not intra-articular fractures.

Hardware: GE 64-row spiral CT machine (Light Speed spiral CT, GE, USA). Dell high-performance computer (CPU: E3-1225 V2 3.20GHz, memory: 16GB, graphics card: NVIDIA Quadro K4200, operating system: Windows 10, 64-bit).

Software: Mimics Research 20.0, 3-matic Research 12.0 (Materialise, Belgium), provided by Sandi Tribe (Shanghai) Technology Co., Ltd.

The lower limbs of the patients were placed in parallel with the toes up, and a full-length scan was performed. The CT scan interval was 0.625mm, and the matrix size was 512512 pixels. The scan voltage was 140kV, the exposure was 100 mAs, and the screw pitch was 0.625mm (GE 64-slice spiral CT machine with automatic tube current control system with the same scan parameters on both sides). The obtained general DICOM 3.0 standard format data were stored.

Mimics 20.0 software was used to directly read the CT images in DICOM 3.0 format. Three-dimensional geometric models were established under the same threshold conditions for the affected side with internal fixation, the affected side without internal fixation and the unaffected side, and the average CT data of the models were recorded.

The three sets of data from the models of the healthy side, the affected side with internal fixation, and the affected side without internal fixation were imported into 3-matic Research 12.0. An adaptive triangle mesh was applied, and the side length was set to 1mm to optimize the details of the model. The detailed characteristics of the model were retained, and the maximum wall thickness threshold was set to 10,000mm during wall thickness analysis.

To interpret the weak areas of the cortical bone in terms of the wall thickness and the transformation in terms of the relationship between the density and stiffness of the material, the median wall thickness of the unaffected limb and the affected limb with and without internal fixation was measured. The analysis was performed as follows: in the Analyze tab, the Create Wall Thickness Analysis button was clicked, and then Cortical As Entity was selected. The threshold was set to 10,000.0mm. A histogram with the wall thickness distribution was displayed, and a series of colors was visualized on the Cortical 3D object, with green representing thinner structures and red corresponding to thicker areas.

We performed three-dimensional reconstruction of the CT data of the healthy and affected segments under the same conditions, resulting in wall thickness graphs corresponding to the basic phase and the target phase, respectively. Three-dimensional reconstruction of the CT data of the affected limb with simulated removal of the internal fixation was performed, yielding a wall thickness graph corresponding to the simulated phase (Fig.1).

Three phases of wall thickness analysis: the basic phase (A), the target phase (B), and the simulated phase (C).

The ratio of the median wall thickness and the average CT value in the simulated phase to the corresponding values in the basic phase was calculated to obtain the ratios R2 and R4: R2=median wall thickness of the simulated phase/median wall thickness of the basic phase; R4=average CT value of the simulated phase/average CT value of the basic phase. The product of the average CT value and the median wall thickness was defined as the healing index (HI), and R5=simulated phase HI/basic phase HI.

The fracture healing state was evaluated through imaging and clinical examinations at 9months after surgery. The bone healing of patients was observed for half a year after removal of the internal fixation to monitor for refracture. In the case that nonunion continued to be observed, bone grafting and internal fixation were performed again. Patients with poor bone healing were further observed. If there were no signs of healing, bone grafting was performed.

The criteria for bone healing were as follows: X-ray images showed blurring of the fracture line and a continuous callus passing through the fracture line11; additionally, upon the release of external fixation, the patient does not have any tenderness at the fracture site, can walk with weight, and has no longitudinal pain in the fractured limb on percussion12.

The criteria for bone nonunion were as follows: Nonunion was defined by pain and abnormal activity at the fracture site, persistent light-transmitting bands on X-ray examination, and no progress in the formation of the callus at 12weeks after treatment13.

Poor bone healing was defined as a healing state between that of bone healing and bone nonunion.

Measurement data are expressed as the meanstandard deviation or median, and Pearson or Spearman correlation analysis was performed. One-way ANOVA was conducted to analyze differences between multiple groups. Receiver operating characteristic (ROC) curve analysis was performed for the diagnostic analysis, and the critical point of diagnosis was analyzed by the maximum Youden index method. P<0.05 was set as statistically significant, and all data analyses were performed by SPSS (version 20.0; IBM Corp., Armonk, NY, USA).

Informed consent was obtained from all the patients and the study was approved by Biomedical Ethics Committee of Anhui Medical University (reference number 20150129). The study has been performed in accordance with the ethical standards of the Declaration of Helsinki in 1964.

Read more here:
Wall thickness analysis method for judging the degree of lower ... - Nature.com