Segmentation of Charcot’s Neuropathy (Diabetic Foot) Using Medical Images and Artificial Intelligence
Segmentation of Charcot’s Neuropathy
Keywords:
Diabetic Foot (DF), Charcot neuropathy, Artificial intelligence, plain radiography, synthetic data, generative adversarial networks, image segmentation, U-NetAbstract
Background: Charcot neuroarthropathy (CN) is a diabetes-associated neuropathic condition that can cause progressive bone and joint destruction. The scarcity of annotated radiographs limits the development and evaluation of deep learning models for radiographic image segmentation.
Objectives: This study developed and evaluated a generative adversarial network (GAN)-augmented deep learning framework for segmentation of the affected bone/joint region in Charcot foot radiographs.
Methodology: Twenty plain radiographs from four patients with CN stages 0-2 were included. Fourteen original radiographs were used to train SinGAN-Seg, which generated 560 synthetic radiographs with corresponding segmentation masks. Synthetic-image similarity was assessed using single-image Fréchet Inception Distance (SIFID) and Inception Score (IS). Six U-Net-based segmentation architectures were trained using the synthetic dataset. Of the 560 synthetic image–mask pairs, 504 (90%) were used for training, and 56 (10%) were reserved for testing. Segmentation performance was evaluated using the intersection over union (IoU).
Results: The SIFID values ranged from 77.56 to 125.81 for images generated from reference image A and from 139.92 to 196.91 for images generated from reference image B. Across the evaluated architectures, mean IoU values ranged from 0.642 to 0.757 on the held-out synthetic test images. R2U-Net obtained the highest observed mean IoU (0.757); however, no formal statistical comparison was performed, and this result should be interpreted cautiously.
Conclusion: GAN-generated image–mask pairs enabled preliminary training and evaluation of U-Net-based models for segmenting the annotated affected bone/joint region in Charcot foot radiographs. Larger patient-level, externally validated datasets are needed before conclusions can be made regarding diagnostic accuracy, clinical reliability, or generalizability.
References
Rosskopf AB, Loupatatzis C, Pfirrmann CWA, Böni T, Berli MC. The Charcot foot: a pictorial review. Insights Imaging. 2019 Dec 5;10(1):77. doi:10.1186/s13244-019-0768-9
Mascio A, Comisi C, Cinelli V, Pitocco D, Greco T, Maccauro G, et al. Radiological Assessment of Charcot Neuro-Osteoarthropathy in Diabetic Foot: A Narrative Review. Diagnostics. 2025 Mar 19;15(6):767. doi:10.3390/diagnostics15060767
Benard T, Lequint C, Jugnet AC, Bouly M, Winther T, Penfornis A, et al. Delayed Diagnosis of Bilateral Neuroarthropathy: Serious Impact on the Development of Charcot’s Foot, a Case Report. Medicina (B Aires). 2022 Nov 30;58(12):1763. doi:10.3390/medicina58121763
Fahmy AM. Assessing the Efficacy of AI Models in Medical Imaging Analysis for the Early Identification of Diabetic Foot Ulcers and Gangrene: A Review. Premier Journal of Science. 2025 Jul 30. doi:10.70389/PJS.100093
Kim J. The pathophysiology of diabetic foot: a narrative review. Journal of Yeungnam Medical Science. 2023 Oct 31;40(4):328–34. doi:10.12701/jyms.2023.00731
Cakir M, Tulum G, Cuce F, Yilmaz KB, Aralasmak A, Isik Mİ, et al. Differential Diagnosis of Diabetic Foot Osteomyelitis and Charcot Neuropathic Osteoarthropathy with Deep Learning Methods. Journal of Imaging Informatics in Medicine. 2024 Mar 15;37(5):2454–65. doi:10.1007/s10278-024-01067-0
Martín Noguerol T, Luna Alcalá A, Beltrán LS, Gómez Cabrera M, Broncano Cabrero J, Vilanova JC. Advanced MR Imaging Techniques for Differentiation of Neuropathic Arthropathy and Osteomyelitis in the Diabetic Foot. RadioGraphics. 2017 Jul;37(4):1161–80. doi:10.1148/rg.2017160101
Álvaro-Afonso FJ, Tardáguila-García A, López-Moral M, Sanz-Corbalán I, García-Morales E, Lázaro-Martínez JL. Using Artificial Intelligence for Detecting Diabetic Foot Osteomyelitis: Validation of Deep Learning Model for Plain Radiograph Interpretation. Applied Sciences. 2025 Aug 1;15(15):8583. doi:10.3390/app15158583
Ronneberger O, Fischer P, Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation. In: Navab, N., Hornegger, J., Wells, W., Frangi, A. (eds) Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015 [Internet]. Springer, Cham; 2015. Available from: https://doi.org/10.1007/978-3-319-24574-4_28 doi:https://doi.org/10.1007/978-3-319-24574-4_28
Renard F, Guedria S, Palma N De, Vuillerme N. Variability and reproducibility in deep learning for medical image segmentation. Sci Rep. 2020 Aug 13;10(1):13724. doi:10.1038/s41598-020-69920-0
Willemink MJ, Koszek WA, Hardell C, Wu J, Fleischmann D, Harvey H, et al. Preparing Medical Imaging Data for Machine Learning. Radiology. 2020 Apr;295(1):4–15. doi:10.1148/radiol.2020192224
Thambawita V, Salehi P, Sheshkal SA, Hicks SA, Hammer HL, Parasa S, et al. SinGAN-Seg: Synthetic training data generation for medical image segmentation. PLoS One. 2022 May 2;17(5):e0267976. doi:10.1371/journal.pone.0267976
Heusel M, Ramsauer H, Unterthiner T, Nessler B, Hochreiter S. GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium. In: Advances in Neural Information Processing Systems. 2017.
Salimans T, Goodfellow I, Zaremba W, Cheung V, Radford A, Chen X. Improved Techniques for Training GANs [Internet]. 2016 Jun 10. Available from: http://arxiv.org/abs/1606.03498
Zhou Z, Siddiquee MMR, Tajbakhsh N, Liang J. UNet++: A Nested U-Net Architecture for Medical Image Segmentation [Internet]. 2018 Jul 18. Available from: http://arxiv.org/abs/1807.10165
Oktay O, Schlemper J, Folgoc L Le, Lee M, Heinrich M, Misawa K, et al. Attention U-Net: Learning Where to Look for the Pancreas [Internet]. 2018 May 20. Available from: http://arxiv.org/abs/1804.03999
Alom MZ, Hasan M, Yakopcic C, Taha TM, Asari VK. Recurrent Residual Convolutional Neural Network based on U-Net (R2U-Net) for Medical Image Segmentation [Internet]. 2018. Available from: https://arxiv.org/abs/1802.06955
Gatys LA, Ecker AS, Bethge M. Image Style Transfer Using Convolutional Neural Networks. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2016. p. 2414–23.
