Automatic Measurement Of Proximal Femoral Parameters Based On Generative Adversarial Networks

With the development of society,China gradually steps into the aging society,followed by a series of diseases of the elderly rising.Femoral fracture accounts for a large proportion in senile diseases,and the treatment is mainly artificial prosthesis replacement.At present,the main source of artificial prosthesis in China is imported from abroad.This will not only make the surgery more complicated,but also affect the postoperative bone recovery.The measurement of proximal femoral parameters can design artificial prosthesis in line with Chinese people,and even achieve personalized artificial prosthesis in the future.On the other hand,the establishment of Chinese femoral parameter database can help researchers in their research work.In China's large population,parameters of proximal femur vary due to various factors.Manual measurement requires a huge amount of manpower and material resources,and there are subjective errors.The existing methods of automatic measurement also have the problem of inaccurate measurement.With the development of deep learning,medical image processing based on deep learning has become a hot topic and many achievements have been made.In this context,we respond to the requests of Xi 'an Honghui hospital and cooperate with them deeply.The study was based on data provided by the Xi 'an Honghui hospital.We propose a method of proximal femoral parameter measurement based on the Generative Adversarial Network(GAN).The main work of this paper is as follows.1.We analyzed the significance and feasibility of the measurement of proximal femoral parameters,and then comprehensively understood the current research status of the measurement of femoral parameters.Combined with the existing automatic measurement,we learned from the desirable experience,and finally summed up the key technology of automatic measurement of proximal femoral parameters.2.We analyzed the morphological characteristics of proximal femoral parameters and studied the methods of physician measurement.In this paper,we demonstrate in detail the realization of several kinds of computer automatic measurement of proximal femoral parameters,and then determine the implementation scheme and form the algorithm flow.Finally,the evaluation index of the algorithm is given.3.Because of the complexity of X-ray images of hip joint,direct measurement of parameters in hip joint will affect the accuracy of parameters measurement of proximal femur.We propose that the femur be accurately separated from the hip joint and then measured.The results of the generated adversarial networks were compared with the results of manual segmentation and other networks.4.On the basis of femoral segmentation,we propose some measurement methods for specific parameters.In this paper,a method combining the GAN algorithm and thecircle fitting algorithm is proposed to determine the position of the femoral ball center and the radius of the femoral head.Corner detection algorithm and edge detection algorithm were used to calculate femoral neck shaft angle(NSA),femoral eccentricity and femoral shaft outer diameter.Then the experimental results are analyzed and evaluated.From the experimental results,the measured results of all parameters are within the range reported in the literature.Compared with the results of manual measurement,the algorithm in this paper is somewhat different,but the gap is within the acceptable range.Various evaluation indexes are used for evaluation,which also proves that the algorithm in this paper has the characteristics of accurate measurement results and stable performance.Considering the current situation of domestic research,this algorithm has certain reference value.