Existing state-of-the-art “safe zone” prediction methods are statistics-based methods, image-matching techniques, and machine learning methods. Yet, those methods bring a tension between accuracy and interpretability.
Purpose
To explore the model explanations and estimator consensus for “safe zone” prediction.
Material and Methods
We collected the pelvic datasets from Orthopaedic Hospital, and a novel acetabular cup detection method is proposed for automatic ROI segmentation. Hybrid priors comprising both specific priors from data and general priors from experts are constructed. Specifically, specific priors are constructed based on the fine-tuned ResNet-101 convolutional neural networks (CNN) model, and general priors are constructed based on expert knowledge. Our method considers the model explanations and dynamic consensus through appending a SHapley Additive exPlanations (SHAP) module and a dynamic estimator stacking.
Results
The proposed method achieves an accuracy of 99.40% and an area under the curve of 0.9998. Experimental results show that our model achieves superior results to the state-of-the-art conventional ensemble classifiers and deep CNN models.
Conclusion
This new screening model provides a new option for the “safe zone” prediction of acetabular cup.
MurphyWSYunHHHaydenB, et al.The safe zone range for cup anteversion is narrower than for inclination in THA. Clin Orthop Rel Res2018;476:325–335.
2.
SadhuANamDCoobsBR, et al.Acetabular component position and the risk of dislocation following primary and revision total hip arthroplasty: a matched cohort analysis. J Arthroplasty2017;32:987–991.
3.
KanawadeVDorrLDBanksSA, et al.Precision of robotic guided instrumentation for acetabular component positioning. J Arthroplasty2015;30:392–397.
4.
LewinnekGELewisJTarrR, et al.Dislocations after total hip-replacement arthroplasties. J Bone Joint Surg Am1978;60:217–220.
5.
DombBGEl BitarYFSadikAY, et al.Comparison of robotic-assisted and conventional acetabular cup placement in THA: a matched-pair controlled study. Clin Orthop Rel Res2014;472:329–336.
6.
HohmannEBryantATetsworthK. A comparison between imageless navigated and manual freehand technique acetabular cup placement in total hip arthroplasty. J Arthroplasty2011;26:1078–1082.
7.
CallananMCJarrettBBragdonCR, et al.The John Charnley Award: risk factors for cup malpositioning: quality improvement through a joint registry at a tertiary hospital. Clin Orthop Rel Res2011;469:319–329.
8.
WidmerK-H. A simplified method to determine acetabular cup anteversion from plain radiographs. J Arthroplasty2004;19:387–390.
9.
HassanDMJohnstonGHDustWN, et al.Radiographic calculation of anteversion in acetabular prostheses. J Arthroplasty1995;10:369–372.
10.
AcklandMBourneWUhthoffH. Anteversion of the acetabular cup. Measurement of angle after total hip replacement. J Bone Joint Surg Br1986;68:409–413.
11.
LiawC-KHouS-MYangR-S, et al.A new tool for measuring cup orientation in total hip arthroplasties from plain radiographs. Clin Orthop Rel Res2006;451:134–139.
12.
VisserJKoningsJG. A new method for measuring angles after total hip arthroplasty. A study of the acetabular cup and femoral component. J Bone Joint Surg Br1981;63:556–559.
13.
PradhanR. Planar anteversion of the acetabular cup as determined from plain anteroposterior radiographs. J Bone Joint Surg Br1999;81:431–435.
14.
KawaharaSHaraTSatoT, et al.Digitalized analyses of intraoperative acetabular component position using image-matching technique in total hip arthroplasty. Bone Jt Res2020;9:360–367.
15.
OnanA. Bidirectional convolutional recurrent neural network architecture with group-wise enhancement mechanism for text sentiment classification. J King Saud Univ Comput Inf Sci2022;34:2098–2117.
16.
OnanA. Sentiment analysis on product reviews based on weighted word embeddings and deep neural networks. Concurr Comput Pract Exp2021;33:e5909.
17.
OnanA. Sentiment analysis on massive open online course evaluations: a text mining and deep learning approach. Comput Appl Eng Educ2021;29:572–589.
18.
OnanA. Mining opinions from instructor evaluation reviews: a deep learning approach. Comput Appl Eng Educ2020;28:117–138.
19.
OnanAKorukoğluSBulutH. A hybrid ensemble pruning approach based on consensus clustering and multi-objective evolutionary algorithm for sentiment classification. Inf Process Manage2017;53:814–833.
20.
OnanAKorukoğluS. A feature selection model based on genetic rank aggregation for text sentiment classification. J Inf Sci2017;43:25–38.
21.
OnanA. Consensus clustering-based undersampling approach to imbalanced learning. Sci Program2019;2019:5901087.
22.
OnanA. Two-stage topic extraction model for bibliometric data analysis based on word embeddings and clustering. IEEE Access2019;7:145614–145633.
23.
OnanAKorukoğluSBulutH. Ensemble of keyword extraction methods and classifiers in text classification. Expert Syst Appl2016;57:232–247.
24.
OnanA. An ensemble scheme based on language function analysis and feature engineering for text genre classification. J Inf Sci2018;44:28–47.
25.
OnanA. Biomedical text categorization based on ensemble pruning and optimized topic modelling. Comput Math Method Med2018;2018:1–22.
26.
OnanAToçoğluMA. A term weighted neural language model and stacked bidirectional LSTM based framework for sarcasm identification. IEEE Access2021;9:7701–7722.
27.
OnanA. Topic-enriched word embeddings for sarcasm identification. In: Proceedings of the computer science on-line conference, 2019 Apr 24–27, 293–304. Zlin, Czech Republic: Springer.
28.
Han F, Liao S, Xiong C, et al. Explainable prediction of whether the acetabular cup is placed in the “safe zone” from X-ray images. In: Proceedings of the IEEE international conference on bioinformatics and biomedicine, 9–12 Dec 2021, 882-889. TX, NJ: IEEE.
29.
LiangYLiSYanC, et al.Explaining the black-box model: a survey of local interpretation methods for deep neural networks. Neurocomputing2021;419:168–182.
30.
GuidottiRMonrealeARuggieriS, et al.A survey of methods for explaining black box models. ACM Comput Surv2019;51:1–42.
31.
ChenJParisSDurandF. Real-time edge-aware image processing with the bilateral grid. ACM Trans Graph2007;26:103.
32.
LuCXiaSShaoM, et al.Arc-support line segments revisited: an efficient high-quality ellipse detection. IEEE Trans Image Process2020;29:768–781.
33.
PratapTKokilP. Computer-aided diagnosis of cataract using deep transfer learning. Biomed Signal Process Control2019;53:101533.
LohWY. Classification and regression trees. Wiley Interdiscip Rev Data Mining Knowl Discov2011;1:14–23.
36.
BreimanL. Random forests. Mach Learn2001;45:5–32.
37.
FriedmanJH. Greedy function approximation: a gradient boosting machine. Ann Stat2001;29(29):1189–1232.
38.
ChenTGuestrinC. Xgboost: a scalable tree boosting system. In: Proceedings of the ACM SIGKDD international conference on knowledge discovery and data mining, 13–17 Aug 2016, 785–794. San Francisco, CA: Association for Computing Machinery.
39.
DhananjayBSivaramanJ. Analysis and classification of heart rate using CatBoost feature ranking model. Biomed Signal Process Control2021;68:102610.
40.
LundbergSMErionGChenH, et al.From local explanations to global understanding with explainable AI for trees. Nat Mach Intell2020;2:56–67.
41.
FidelGBittonRShabtaiA. When explainability meets adversarial learning: Detecting adversarial examples using SHAP signatures. In: Proceedings of International Joint Conference on Neural Networks; 19–24 July 2020, 1–8. Glasgow, NJ: IEEE.
42.
LundbergSMLeeS-I. A unified approach to interpreting model predictions. In: Proceedings of the international conference on neural information processing systems, 4–9 Dec 2017, 4768–4777. Long Beach, CA: Curran Associates Inc.
43.
RibeiroMTSinghSGuestrinC. “Why should I trust you?”: explaining the predictions of any classifier. In: Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016 Aug 13–17, pp.1135–1144. San Francisco, CA: Association for Computing Machinery.
44.
ShapleyLS. A value for n-person games. In: KuhnHWTuckerAW, editors.Annals of mathematics studies. Princeton, NJ: Princeton University Press, 1953:31–40.
45.
LundbergSMErionGGLeeS-I. Consistent individualized feature attribution for tree ensembles. arXiv Preprint2018: 180203888.
46.
HuangGLiuZMaatenLVD, et al. Densely connected convolutional networks. In: Proceedings of the IEEE conference on computer vision and pattern recognition, 21–26 July 2017, 2261–2269. Honolulu, HI:, IEEE.
47.
SzegedyCLiuWJiaY, et al. Going deeper with convolutions. In: Proceedings of the IEEE conference on computer vision and pattern recognition, 2015 Jun 7–12, 1–9. Boston, MA: IEEE.
48.
SzegedyCVanhouckeVIoffeS, et al.Rethinking the inception architecture for computer vision. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016 Jun 27–30, pp.2818–2826. Las Vegas, NV: IEEE.
49.
HeKZhangXRenS, et al.Deep residual learning for image recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016 Jun 27–30, pp.770–778. Las Vegas, NV: IEEE.
50.
ZhangXZhouXLinM, et al. Shufflenet: an extremely efficient convolutional neural network for mobile devices. In: Proceedings of the IEEE conference on computer vision and pattern recognition, 18–23 Jun 2018, 6848–6856; UT. NJ. IEEE.
51.
CholletF. Xception: Deep learning with depthwise separable convolutions. In: Proceedings of the IEEE conference on computer vision and pattern recognition, 21–26 Jul 2017, 1800–1807. HI. NJ. IEEE.
52.
MemişAVarlıSBilgiliF. Image based quantification of the proximal femur shape deformities in 3D by using the contralateral healthy shape structure: a preliminary study. Biomed Signal Process Control2022;71:103079.
