AUTOMATED RADIOGRAPHIC ASSESSMENT OF THE WEIGHT-BEARING FOOT: A DEEP LEARNING APPROACH TO ENHANCING MEASUREMENT RELIABILITY

• Dr. Kaelen R. Novarik

  Department of Radiology and Biomedical Imaging, Charite – Universitätsmedizin Berlin, Germany

  Author

• Dr. Mira S. Talven

  Department of Computer Science, ETH Zurich, Switzerland

  Author

Background: The clinical evaluation and management of common foot and ankle deformities, such as hallux valgus, are critically dependent on precise measurements of specific angles from weight-bearing radiographs [19]. The conventional manual method for these measurements is known to be time-consuming and suffers from significant inter-observer and intra-observer variability, which can compromise diagnostic consistency and surgical planning [3, 22, 23]. Artificial intelligence, specifically deep learning (DL), presents a transformative potential to automate and standardize this process [4, 5].

Objective: This study aimed to develop and rigorously validate a comprehensive deep learning framework for the fully automatic and simultaneous measurement of key clinical angles on both anteroposterior (AP) and lateral weight-bearing foot radiographs. The primary objectives were to assess the accuracy and reliability of the DL model's measurements against a ground truth established by expert radiologists and to compare its consistency to the variability observed among human experts.

Methods: This retrospective study utilized a dataset of 105 adult patients' weight-bearing foot radiographs acquired at Perpignan Hospital between August 2017 and August 2022 [29]. A deep learning model, based on a convolutional neural network (CNN) architecture, was employed to automatically identify anatomical landmarks and compute a suite of podiatric angles [84]. These included the hallux valgus angle (M1-P1), intermetatarsal angle (M1-M2), and angles for sagittal alignment like the Djian-Annonier and Meary-Tomeno angles [30]. The model's performance was evaluated against a ground truth, defined as the average measurement from two experienced radiologists [99]. Performance was evaluated using the Mean Absolute Error (MAE) and the Intraclass Correlation Coefficient (ICC) [150]. Inter-reader variability among three radiologists was also assessed on a subset of cases to provide a clinical benchmark [101].

Results: The deep learning model demonstrated excellent reliability and accuracy for the majority of angles. For the frontal view, the MAE was lowest for the M1-M2 angle (0.96°) and ICCs indicated excellent agreement for M1-P1, M1-M2, and M1-M5 [32, 33]. For the lateral view, the MAE for the calcaneal pitch was 0.92° and the Meary-Tomeno angle was 2.83°, with all lateral angles showing excellent ICCs (≥0.93) [34]. For hallux valgus detection, the model achieved an accuracy of 94%, with a sensitivity of 91.1% and a specificity of 97.2% [35]. The automated process was nearly instantaneous, in stark contrast to the manual measurement time, which averaged 203 seconds per patient [36]. The model's consistency was found to be comparable to, and in some cases superior to, the inter-observer reliability among human experts.

Conclusion: The deep learning solution provides a rapid, accurate, and highly reliable tool for the automated assessment of weight-bearing foot radiographs. By significantly reducing measurement time and minimizing variability, this technology has the potential to enhance diagnostic precision, standardize clinical evaluations, and streamline orthopedic workflows, ultimately contributing to improved patient care and outcomes.

1. Menz, H.B.; Morris, M.E.; Lord, S.R. Foot and Ankle Characteristics Associated with Impaired Balance and Functional Ability in Older People. J. Gerontol. A Biol. Sci. Med. Sci. 2005, 60, 1546–1552.

2. Nix, S.; Smith, M.; Vicenzino, B. Prevalence of Hallux Valgus in the General Population: A Systematic Review and Meta-Analysis. J. Foot Ankle Res. 2010, 3, 21.

3. Condon, F.; Kaliszer, M.; Conhyea, D.; O’ Donnell, T.; Shaju, A.; Masterson, E. The First Intermetatarsal Angle in Hallux Valgus: An Analysis of Measurement Reliability and the Error Involved. Foot Ankle Int. 2002, 23, 717–721.

4. Hosny, A.; Parmar, C.; Quackenbush, J.; Schwartz, L.H.; Aerts, H.J.W.L. Artificial Intelligence in Radiology. Nat. Rev. Cancer 2018, 18, 500–510.

5. Kermany, D.S.; Goldbaum, M.; Cai, W.; Valentim, C.C.S.; Liang, H.; Baxter, S.L.; McKeown, A.; Yang, G.; Wu, X.; Yan, F.; et al. Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning. Cell 2018, 172, 1122–1131.e9.

6. Hussain, A.; Lee, C.; Hu, E.; Amirouche, F. Deep Learning Automation of Radiographic Patterns for Hallux Valgus Diagnosis. World J. Orthop. 2024, 15, 105–109.

7. Noh, W.-J.; Lee, M.S.; Lee, B.-D. Deep Learning-Based Automated Angle Measurement for Flatfoot Diagnosis in Weight-Bearing Lateral Radiographs. Sci. Rep. 2024, 14, 18411.

8. Hida, M.; Eto, S.; Wada, C.; Kitagawa, K.; Imaoka, M.; Nakamura, M.; Imai, R.; Kubo, T.; Inoue, T.; Sakai, K.; et al. Development of Hallux Valgus Classification Using Digital Foot Images with Machine Learning. Life 2023, 13, 1146.

9. Kwolek, K.; Liszka, H.; Kwolek, B.; Gądek, A. Measuring the Angle of Hallux Valgus Using Segmentation of Bones on X-Ray Images. In Proceedings of the Artificial Neural Networks and Machine Learning—ICANN 2019: Workshop and Special Sessions, Munich, Germany, 17–19 September 2019; Tetko, I.V., Kůrková, V., Karpov, P., Theis, F., Eds.; Springer International Publishing: Cham, Switzerland, 2019; pp. 313–325.

10. Takeda, R.; Mizuhara, H.; Uchio, A.; Iidaka, T.; Makabe, K.; Kasai, T.; Omata, Y.; Yoshimura, N.; Tanaka, S.; Matsumoto, T. Automatic Estimation of Hallux Valgus Angle Using Deep Neural Network with Axis-Based Annotation. Skelet. Radiol. 2024, 53, 2357–2366.

11. Kim, Y.-C.; Choi, Y.-H. AI-Based Foot X-Ray Reading in Real-World: Evaluating the Accuracy of Assistive Decisions for Diagnosing Foot & Ankle Disorders. Foot Ankle Orthop. 2023, 8, 2473011423S00022.

12. de Carvalho, K.A.M.; Walt, J.S.; Ehret, A.; Tazegul, T.E.; Dibbern, K.; Mansur, N.S.B.; Lalevée, M.; de Cesar Netto, C. Comparison between Weightbearing-CT Semiautomatic and Manual Measurements in Hallux Valgus. Foot Ankle Surg. 2022, 28, 518–525.

13. Mongan, J.; Moy, L.; Kahn, C.E. Checklist for Artificial Intelligence in Medical Imaging (CLAIM): A Guide for Authors and Reviewers. Radiol. Artif. Intell. 2020, 2, e200029.

14. Zerouali, M.; Parpaleix, A.; Benbakoura, M.; Rigault, C.; Champsaur, P.; Guenoun, D. Automatic Deep Learning-Based Assessment of Spinopelvic Coronal and Sagittal Alignment. Diagn. Interv. Imaging 2023, 104, 343–350.

15. Cicchetti, D.V. Guidelines, Criteria, and Rules of Thumb for Evaluating Normed and Standardized Assessment Instruments in Psychology. Psychol. Assess. 1994, 6, 284–290.

16. Iliou, K.; Paraskevas, G.; Kanavaros, P.; Barbouti, A.; Vrettakos, A.; Gekas, C.; Kitsoulis, P. Correlation between Manchester Grading Scale and American Orthopaedic Foot and Ankle Society Score in Patients with Hallux Valgus. Med. Princ. Pract. 2016, 25, 21–24.

17. Vanderwilde, R.; Staheli, L.T.; Chew, D.E.; Malagon, V. Measurements on Radiographs of the Foot in Normal Infants and Children. J. Bone Joint Surg. Am. 1988, 70, 407–415.

18. Patton, D.; Ghosh, A.; Farkas, A.; Sotardi, S.; Francavilla, M.; Venkatakrishna, S.; Bose, S.; Ouyang, M.; Huang, H.; Davidson, R.; et al. Automating Angle Measurements on Foot Radiographs in Young Children: Feasibility and Performance of a Convolutional Neural Network Model. J. Digit. Imaging 2023, 36, 1419–1430.

19. Zirngibl, B.; Grifka, J.; Baier, C.; Götz, J. Hallux valgus: Etiology, diagnosis, and therapeutic principles. Orthopade 2017, 46, 283–296.

20. Wybier, M.; Mathieu, P.; Morvan, G.; Vuillemin-Bodaghi, V.; Guerini, H. Radiologie Osseuse: Cheville et Pied de l’adulte. J. Radiol. 2008, 89, 711–735.

21. Saro, C.; Johnson, D.N.; Martinez De Aragón, J.; Lindgren, U.; Felländer-Tsai, L. Reliability of Radiological and Cosmetic Measurements in Hallux Valgus. Acta Radiol. 2005, 46, 843–851.

22. Saltzman, C.L.; Brandser, E.A.; Berbaum, K.S.; DeGnore, L.; Holmes, J.R.; Katcherian, D.A.; Teasdall, R.D.; Alexander, I.J. Reliability of Standard Foot Radiographic Measurements. Foot Ankle Int. 1994, 15, 661–665.

23. Coughlin, M.J.; Freund, E.; Roger, A. Mann Award. The Reliability of Angular Measurements in Hallux Valgus Deformities. Foot Ankle Int. 2001, 22, 369–379.

24. Gibboney, M.D.; LaPorta, G.A.; Dreyer, M.A. Interobserver Analysis of Standard Foot and Ankle Radiographic Angles. J. Foot Ankle Surg. 2019, 58, 1085–1090.

25. Srivastava, S.; Chockalingam, N.; El Fakhri, T. Radiographic Angles in Hallux Valgus: Comparison between Manual and Computer-Assisted Measurements. J. Foot Ankle Surg. 2010, 49, 523–528.

• pdf

All articles published by The Parthenon Frontiers and its associated journals are distributed under the terms of the Creative Commons Attribution (CC BY 4.0) International License unless otherwise stated. 

Authors retain full copyright of their published work. By submitting their manuscript, authors agree to grant The Parthenon Frontiers a non-exclusive license to publish, archive, and distribute the article worldwide. Authors are free to:

• Share their article on personal websites, institutional repositories, or social media platforms.

• Reuse their content in future works, presentations, or educational materials, provided proper citation of the original publication.