Application Study Of Digital Orthopedic Technology In The Diagnosis And Treatment Of Acetabular Fractures

Objective: To develop and test a deep learning model for detecting acetabular fractures on pelvic anteroposterior radiographs(PARs)and compare its performance to that of clinicians.Methods: We retrospectively enrolled 1120 patients with acetabular fracture in our hospital and divided the fracture into types A,B and C according to the Three-Column Classification theory of acetabular fracture.The PAR of the patient was collected from the PACS and was randomly allocated at a 3:1 ratio for the deep learning model’s training and internal test.A total of 86 patients were enrolled from another two independent tertiary hospitals for external validation.The X-ray film was preprocessed,including the region of interest(ROI)cropping and flipping,image annotation,data augmentation,etc.,and was then input to the deep learning model for identifying acetabular fractures based on Dense Convolution Network(Dense Net)architecture.After that,the deep learning model was trained,tested and verified.The output results are supported by Gradient-weighted class activation mapping(Grad-CAM)technology.In addition,10 orthopedic trauma clinicians of our hospital were recruited for the reading test of identifying acetabular fractures.The potential misdiagnosed case(PMC)was determined according to the test results.The sensitivity,specificity and accuracy of clinicians and the deep learning model in detecting acetabular fractures were statistically analyzed and compared.The area under the ROC curve(AUC)was used to evaluate the ability of the deep learning model to detect different fracture subtypes.Results: The means of 10 clinicians’ sensitivity,specificity,and accuracy to identify acetabular fractures were 0.750/0.735,0.909/0.909,and 0.829/0.822,in the internal test/external validation set,respectively.The sensitivity,specificity,and accuracy of the deep learning model were 0.926/0.872,0.978/0.988,and 0.952/0.930,respectively.The deep learning model identified type A fractures with an AUC of 0.963(95% confidence interval(CI): 0.927–0.985)/0.950(95% CI: 0.867–0.989);type B fractures with an AUC of 0.991(95% CI: 0.967–0.999)/0.989(95% CI: 0.930–1.000);and type C fractures with an AUC of 1.000(95% CI:0.975–1.000)/1.000(95% CI: 0.897–1.000)in the test/validation set.The deep learning model correctly recognized 56.5%(26/46)of PMCs.Conclusions: A deep learning model for detecting acetabular fractures on PARs is feasible.The results of the deep learning model in this study show that its diagnostic performance is comparable to that of clinicians and even better than some clinicians.This model has excellent detection ability for severe Ctype acetabular fractures and has a certain detection ability for acetabular fractures that clinicians tend to misdiagnose.Objective: T-shaped acetabular fractures are named due to their "T"-like radiographic appearance,but the true fracture line geometry and distribution are not yet known,and the choice of surgical approach is also full of controversy.In this study,we investigated the morphological pattern of T-shaped acetabular fractures by mapping the fracture line using Mimics software.Methods: The clinical data of 1298 patients with acetabular fractures in two big trauma centers were reviewed retrospectively,and a total of 56 patients with T-shaped acetabular fractures were screened out after classification by three doctors.The CT DICOM data of the patient were imported into the Mimics software to perform three-dimensional reconstruction of each independent fracture fragment and virtually reduce the fracture.A healthy half pelvis was selected as a standard template,and the fracture lines of all patients were drawn and superimposed on the template.Frequency heatmaps were generated using E-3D software based on differences in fracture line density distribution.All cases were divided into three groups regarding the distribution characteristics of the longitudinal fracture line: the posterior group,the anterior group,and the extra-posterior group.In patients undergoing surgical treatment,the surgical approach and fixation method were recorded,and the quality of fracture reduction was assessed.Results: The overall fracture line of T-shaped fractures is not shaped like a "T".The transverse fracture line is meandering,but does not extend above the acetabular roof;the longitudinal fracture line is inclined forward or backward along the edge of the acetabular fossa and splits the obturator foramen.The anterior and posterior edges of the acetabular fossa,the transition zone between the acetabular roof and the anterior wall,the transition zone between the superior pubic ramus and the anterior wall,and the posterior wall near the greater sciatic notch are prone to fractures.The shape of T-shaped fractures in different groups was different.The posterior group included 26(46.4%)patients,and the distribution of fracture lines resembled "T" or "Y";the anterior group included 25(44.6%)patients,and the distribution of fracture lines resembled "7";the extra-posterior group included 5(8.9%)patients with a fracture line splitting the ischial body and the obturator intact.Surgical strategies were inconsistent in different groups of T-shaped fracture patients.Fifteen(26.8%)patients had posterior wall fractures,and T-shaped fractures had larger posterior wall fragments.Conclusions: After the occurrence of T-shaped acetabular fractures,there are mainly two fracture lines,horizontal and vertical,sometimes accompanied by fragments of the posterior wall.The distribution of fracture lines of T-shaped fractures has certain rules,and the transition of anatomical structure is the area where fractures are prone to occur.After grouping T-shaped fractures according to the shape of the fracture line,the characteristics of each group are more obvious,which can provide a reference for surgical treatment decisions.Objective: Both-column acetabular fractures are the most common complex acetabular fractures.Although some studies have mentioned the displacement characteristics of both-column fractures,they have not been confirmed.In this study,by simulating the reduction of double-column fractures in computer software,and recording the changes of calibration points and rotation angles before and after reduction,quantitative and qualitative analysis of the spatial displacement characteristics of double-column fractures was conducted.Methods: Eighty-one patients with both-column acetabular fractures from the Third Hospital of Hebei Medical University were retrospectively enrolled.The CT DICOM data of the patient were imported into the Mimics software,and each independent fracture fragment was reconstructed and adjusted to a standard body position.After defining 4 calibration points and 1 rotation angle on the anterior and posterior column fracture fragments,the fracture reduction was simulated based on the mirror reduced template of the healthy side.The coordinate change of the calibration points before and after reduction was recorded,and the degree of the rotation angle was measured at the same time.The displacement distance and direction were analyzed by the Mann-Whitney U test and binomial test,respectively.Results: The anterior column fracture fragments were displaced inwardly and upwardly,with the largest displacement posteriorly inferiorly(P < 0.05),with a median distance of 12.58 [Inter-quartile Range(IQR),9.95] mm.Anterior column fragments were also externally rotated(P < 0.001),with a median rotation angle of 7.93(IQR,10.40)°.The fracture fragments of the posterior column also displaced inward and upward(P < 0.001),and the displacement of the proximal fracture was greater than that of the distal fracture(P < 0.001).Fragments of the posterior column were associated with a significant tendency to internal rotation(P < 0.001),with a median rotation angle of 3.81(IQR,7.49)°.The displacement of anterior column was greater(P < 0.05)in the region of the distal fracture ends of the anterior and posterior column fragments.The angle of rotation of the anterior column fragments was greater than that of the posterior column fragments(P < 0.001).Conclusions: After the both-column acetabular fracture occurs,the displacement and rotation of the fracture fragment can be followed regularly.The fracture fragments of the anterior and posterior columns are displaced inward and upward,the fracture fragments of the anterior column are accompanied by external rotation,and the fracture fragments of the posterior column are accompanied by internal rotation.Fully understanding the spatial displacement characteristics of both-column fractures can help orthopedic surgeons make more efficient surgical decisions.