Research On Measurement Method Of Spinal X-ray Cobb Angle Based On B-spline Curve

ObjectiveThis study intends to develop a method of high-precision recognition and measurement of Cobb Angle in spinal X-ray based on B-spline curve,so as to identify and measure spinal X-ray,which can calculate relatively accurate Cobb Angle for X-ray films similar to those in the Third Hospital of Bethune,Jilin University,with low resolution.At the same time,it provides a set of objective automatic measurement method for orthopedic surgeons,which not only improves the accuracy and reliability of Cobb Angle measurement,but also helps clinicians to collect image data,obtain data and provide auxiliary clinical diagnosis.In this paper,a new model is proposed,which is based on the B-spline spine centerline method to calculate Cobb Angle.This model can calculate Cobb Angle of fuzzy image more accurately.MethodA total of 570 full-length spinal radiographs between 8 and 30 years old were collected from the imaging department of Jilin University Bethune Third Hospital during 2019-2022.The ratio of 6:2:2 was used as the training set,validation set and test set,respectively,to train,verify and test the designed model.The Cobb Angle measured by artificial intelligence model was compared with that measured by professional physicians,and the accuracy and feasibility of the Cobb Angle measurement method based on B-spline curve was confirmed.ResultSPSS26.0 software was used for statistical analysis of the data,P < 0.05 was statistically significant.The difference between the Cobb Angle obtained by the AI program and the Cobb Angle measured by the physician was 1.6380±4.2546,and the intragroup correlation coefficient(ICC)was 0.938 with P<0.001,indicating a high consistency.As for the detection of scoliosis(Cobb Angle ≥10°),Mc Nemar test results showed that P=1 had no significant difference,and Kappa value was 0.828 and P<0.05,indicating a strong consistency between the two detection methods.Further,the diagnostic performance of the model was analyzed by the ROC curve.The area under ROC curve(AUC)of the model was 0.940,the accuracy was 99.13%,the specificity was 99.1%,and the sensitivity was 98.9%,indicating that the model was highly consistent with professional doctors in diagnosis.Four different methods were used to measure the Cobb Angle between the open data set and the 114 test sets.MAE and SMAPE values were calculated and compared respectively.In the open data set,the SMAPE value obtained by the model designed by us ranked second,and MAE ranked first.In the data set provided by the Third Hospital of Bethune,Jilin University,the model designed by us ranks first in SMAPE and first in MAE.It can be seen that the Cobb Angle measured by our designed model is closest to the standard value.ConclusionWe developed a method of Cobb Angle measurement based on B-spline center line of spine,which can automatically and accurately measure Cobb Angle and has the following advantages:(1)It can be used to measure Cobb Angle in fuzzy images;(2)high precision measurement;(3)Rapid measurement time(<1s).Through further research and improvement,the final realization of clinical application.